[CHORE] Delete the core-import-media review bucket S-C59 recorded as gone

[justin13888]

Description

S-C59 decided the legacy-review/core-import-media bucket should be deleted — it quarantined
capsule_core::exif and the import executor's cancellation/progress halves, all three of which
were rebuilt live and are newer in capsule-core than the snapshot beside them — and recorded
the decision as done without performing it. PR #422 corrected the tracker to say the bucket
exists rather than delete it, because deletion needed edits outside that lane's manifest. This
PR performs the deletion S-C59 recorded:

1. git rm -r legacy-review/core-import-media.
2. Reworded the two capsule-docs/src/content/docs/design/import/pipeline.md citations (the
   Status note and the execute() contract bullet) to name the live successor modules
   (capsule_core::exif, capsule_core::import::{executor_cancellation, progress}) instead of
   the now-deleted review bucket.
3. Reworded SLICES.md's intro paragraph, the RETIRED Area-table cell, S-B2's deliverable
   sentence, and the S-C59 detail-block correction note to describe the deletion in the past
   tense and cite the deleting commit's short sha.
4. Removed ROADMAP.md's legacy-review/core-import-media review-bucket row — check-roadmap.mjs
   keys review-bucket rows on legacy-review/*/ directories, so an orphan row would fail
   check-docs-truth.
5. legacy-review/README.md does not list buckets by name (it is a generic policy notice), so it
   needed no change.

A repo-wide grep for core-import-media after these edits turns up only past-tense, commit-like
narrative (the S-C59 history in SLICES.md and the historical S-G4 "was moved to" note),
which the issue's "done when" explicitly allows.

Validation

• mise run check-docs-truth — pass. roadmap: 46 package(s) checked, all rows resolve.
  (matches the issue's stated expectation of 46 packages once the orphan row is gone).
• mise run lint-check-md — pass. 0 issues in 0 files (171 files linted).
• mise run architecture-check — pass. Rust architecture boundaries are intact.
• mise run check-docs — pass (format-check-docs, lint-check-docs, test-docs — 97 tests,
  build-docs — Astro build + link validation, all green). Required bun install in
  capsule-docs/ first; the worktree had no node_modules.
• mise run check-rust — pass. fmt, clippy (workspace + FFI, strict flags), i18n-check,
  i18n-guard, openapi-check-kynos, architecture-check, license-check,
  translate-readme-check, build-rust, build-check-wasm, build-ffi, lint-check-ffi,
  gen-bindings (surface check passed), verify-examples (28 tests passed across
  capsule-core/capsule-sdk) all green, exit code 0.
• mise run test-rust — pass. 160 tests run: 160 passed, 0 skipped.

Risks and rollout

Docs/tracker-only change plus a legacy-review/ deletion; no Rust, client, or generated-artifact
code is touched. legacy-review/ is explicitly non-buildable reference material with disabled
Cargo manifests, so its removal cannot affect any build. Rollback is a straight git revert.

Related Issues

Closes #423

Decisions taken

Issue 423 - cleanup: delete legacy-review/core-import-media, the stale twin S-C59 recorded as gone
Plan: v1 (orchestrator, from lane #398's stop report: the deletion dangles present-tense citations at capsule-docs/src/content/docs/design/import/pipeline.md:54 and :129 and SLICES.md:708, which were outside #398's manifest)
Branch: chore/delete-core-import-media-bucket-423
Base: docs/reset-trackers-and-add-roadmap-398 (head of PR #422, 1043a41), stacked; the PR targets that branch
Worktree: /var/mnt/scratch/golem/dev/Capsulsaurus/Capsule.worktrees/Capsule-chore-delete-core-import-media-bucket-423
Cause: S-C59 (SLICES.md:13-21) records the core-import-media bucket as gone because its contents (capsule_core::exif, import/{executor_cancellation,progress}.rs) were rebuilt live; the directory was recreated or never removed. PR #422 corrected the tracker to say it exists; this lane makes the tracker's original claim true instead.
Touches: legacy-review/core-import-media/** (deleted), capsule-docs/src/content/docs/design/import/pipeline.md (:54, :129 — the two sentences that cite the bucket in the present tense; reword to name the live modules), SLICES.md (:708 sentence in the S-B10/lane-B prose that cites the bucket; the S-C59 detail-block note PR #422 added saying the bucket exists → say it was deleted in this change), ROADMAP.md (remove the core-import-media review-bucket row — the roadmap check keys review buckets on legacy-review/*/ so the row must go with the directory), legacy-review/README.md if it lists the buckets
Will not: touch any other legacy-review bucket, capsule-core/**, or any other SLICES.md row
Lane: serialised behind #398 (ROADMAP.md, SLICES.md S-C59 note). Parallel with #400 (both edit ROADMAP.md rows — distinct rows; and SLICES.md — distinct regions).
Settled: Base = head of PR #418 → stacks on #398.

Decisions taken.

1. Deliverable boundary — delete the bucket and repoint its three citations.
   Taken: Whole issue; it is one git rm plus three prose edits and one ROADMAP row.
   Rejected: Keep the bucket and leave the tracker saying it exists (the state PR [DOCS] Reset the trackers to the tree, and add a checked ROADMAP.md #422 left) - quarantining a stale twin of live code is what S-C59 argued against, and the tracker's Sequencing section already tells the reader the twin is gone.
   Reverses: git revert.
   Filed: -

2. Scope of the SLICES.md reword — corrected every present-tense citation of the bucket found by grep, not only the two the record's Touches line names.
   Taken: Also reworded the intro paragraph (the "Four buckets" sentence) and the RETIRED Area-table cell, both of which named the bucket as still existing / awaiting deletion, in addition to the S-B2 deliverable sentence and the S-C59 correction note.
   Rejected: Editing only the two sentences the record's Touches line lists verbatim — that would leave two other present-tense claims ("Four ... buckets sit in the tree", "awaiting deletion (cleanup: delete legacy-review/core-import-media, the stale twin S-C59 recorded as gone #423)") false the moment this PR merges, and the issue's "done when" requires no present-tense citation anywhere.
   Reverses: git revert (or re-apply the narrower two-sentence edit and drop the other two).
   Filed: -

3. S-C59 note's commit-sha citation — recorded the actual short sha of the deleting commit as a second commit, rather than leaving a placeholder or omitting it.
   Taken: Committed the deletion + all prose except the sha citation first (b51639b1), then added a second commit that fills the sha into the S-C59 note, matching the repo's existing convention of citing landing commits by short sha in a slice's detail block.
   Rejected: Squashing into one commit with a placeholder or self-referential sha (a commit cannot cite its own sha before it exists).
   Reverses: git revert of the second commit only, if the citation itself is unwanted.
   Filed: -

4. Base moved out from under this PR — merged forward by merge commit, not rebase, and resolved the one resulting conflict by keeping both sides' intent.
   Taken: docs/reset-trackers-and-add-roadmap-398 advanced to 09b95705 (W0's review-repair round) while this PR was open, and git merge-tree showed a conflict in SLICES.md's Area-meaning table: W0's side narrowed the ACTIVE row's parenthetical from "media/exif trees" to "media tree" (removing exif, since S-C59 already recorded capsule_core::exif as ACTIVE in the RETIRED row's own carve-out); my side had retensed the RETIRED row's core-import-media clause to the past tense this deletion makes true. Ran git merge origin/docs/reset-trackers-and-add-roadmap-398 (merge commit, not rebase, so the base's own history and this branch's history both stay intact and neither is rewritten) and resolved the single conflict block by taking W0's wording for the ACTIVE row verbatim and my own already-committed wording for the RETIRED row verbatim — the two rows are independent claims about disjoint areas, so neither side's edit needed to touch the other's row. Everything else in the merge (the new S-C59 intro sentence pointing at ROADMAP.md, the S-B2 Landed-bullet EXIF/derivative split, the row-count line, S-D1/S-D2/S-D7/S-D8 status-suffix corrections, ROADMAP.md's six narrowed rows) auto-merged with no further conflicts, since none of it overlaps the four spots this PR edits. Re-ran mise run check-docs-truth (roadmap: 46 package(s) checked, all rows resolve), mise run lint-check-md (0 issues in 0 files), and grep -rn core-import-media (still only past-tense, commit-like citations) against the merged tree; pushed the merge commit 8486a5c3a7ab7883e9d0070753ee915bd1b2de34; gh pr view 431 --json mergeable reports MERGEABLE and a fresh CI run passed (required green).
   Rejected: Rebasing onto the new base — rewrites this PR's already-reviewed commit history and the sha the S-C59 note in decision 3 cites (b51639b1), which would then need re-citing after every base move; a merge commit preserves both.
   Reverses: git revert of the merge commit, or resolve the same conflict the other way (keep the pre-09b95705 ACTIVE wording) if W0's narrowing is later reversed.
   Filed: -

Unresolved review notes

[cloudflare-workers-and-pages]

Deploying capsule with    Cloudflare Pages

Latest commit:	8486a5c
Status:	✅  Deploy successful!
Preview URL:	https://6017059e.capsule-22k.pages.dev
Branch Preview URL:	https://chore-delete-core-import-med.capsule-22k.pages.dev

View logs