fix(teslemetry): correct _tesla_dow - Tesla's day-of-week is Monday=0, not Sunday=0

[springfall2008]

Summary

• _tesla_dow() mapped Python's weekday() (Monday=0) to a wrongly-assumed Tesla convention of Sunday=0, via (python_weekday + 1) % 7. Tesla's tariff_content_v2 fromDayOfWeek/toDayOfWeek actually use the same Monday=0 convention as datetime.weekday().
• Every ON_PEAK boost band (and the day's real-tier layout) therefore landed one day late. During the actual export window the Powerwall saw only the ordinary off-peak tariff, so it never had a price reason to export - it just covered house load.
• This explains every symptom chased in Teslemetry: Powerwall never exports — tariff pushed without optimization_strategy, so TBC stays in Balanced #4600: load-following with grid=0 during "Exporting", raising the boost price changing nothing (it was written to the wrong day), an overnight window importing at the cheap rate while sitting at 97% SoC. optimization_strategy was a red herring - this is the actual cause.
• Fix: make _tesla_dow the identity function - Python's weekday() already matches Tesla's convention.

Testing

• The existing day-of-week tests were self-referential (today_dow = api._tesla_dow(api.base.now.weekday())), so they passed under any mapping and never caught this. Fixed to pin absolute expected indices, computed independently of _tesla_dow (bypassing it entirely, e.g. api.base.now.weekday() / api._local_today_weekday() directly).
• Replaced test_teslemetry_tesla_dow_sunday_zero (which asserted the wrong Sunday=0 behaviour) with test_teslemetry_tesla_dow_matches_python_weekday.
• Added test_teslemetry_build_tariff_boost_resolves_at_the_real_tesla_day_index: a resolver-style regression that independently resolves the built tariff's price at a moment inside the boost window using Tesla's real day convention, and asserts the boosted ON_PEAK price applies - the property that actually matters, and the one that would have caught this.
• ./run_all --quick and ./run_pre_commit both pass.
• Confirmed live (see linked issue): correcting only the day index took a Powerwall from grid=0 to grid=-5156 (full 5kW export) within 50 seconds, no other change.

Fixes #4610

Test plan

• ./run_all --quick
• ./run_pre_commit
• Live confirmation on a real Powerwall (see issue Teslemetry: forced export never works — _tesla_dow writes the boost band one day late (Tesla uses Monday=0) #4610)

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR fixes Teslemetry’s Tesla tariff day-of-week mapping so that fromDayOfWeek/toDayOfWeek align with Tesla’s Monday=0 convention (matching datetime.weekday()), preventing export/boost windows from being written one day late.

Changes:

• Updated TeslemetryAPI._tesla_dow() to be an identity mapping (Python weekday already matches Tesla’s convention).
• Refactored existing tests to avoid self-referential expectations derived from _tesla_dow().
• Added targeted regression tests to ensure boost tiers resolve on the correct Tesla day index.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
apps/predbat/teslemetry.py	Corrects Tesla DOW mapping and documents the convention explicitly.
apps/predbat/tests/test_teslemetry.py	Makes DOW assertions independent of _tesla_dow() and adds resolver-style regression coverage for the boost window.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.