bugfix: simulator crashed, then hung, on Destroy Seed

[hitechhayekian]

Destroy Seed could not complete in the unix simulator. Two separate problems, one behind the other.

1. pyb.Flash was missing (crash)

wipe_flash_filesystem() drives internal flash as a block device via pyb.Flash(start=0) (shared/files.py:59), but unix/variant/pyb.py never defined Flash:

File "shared/actions.py", line 609, in clear_seed
File "shared/files.py", line 59, in wipe_flash_filesystem
AttributeError: 'module' object has no attribute 'Flash'

This fails after both confirmations and after trick PINs are cleared, so the wallet is left half-wiped.

Adds a Flash block-device stub alongside the existing SDCard one, using the same "pretend to work" approach: 512-byte sectors, as files.py:61 asserts, and writes that go nowhere. Also adds testing/test_sim_pyb.py, which holds the simulator's pyb to the attributes shared/ actually references — that is what would have caught this before it reached a menu action.

2. The callgate then hung the simulator

With the crash fixed, Destroy Seed ran to completion and sat on "Clearing..." forever. ckcc.oneway() printed a NOTE and then looped on utime.sleep(60), so the last drawn frame stayed on screen — a finished wipe and a crash look identical.

It now raises SystemExit and names what the bootloader would have done. All four one-way methods end firmware execution on real hardware (DFU, logout, wipe, brick), so all four stop rather than spin. machine.py already does exactly this for bootloader()/soft_reset()/reset(), and the simulator runs with -i, so this lands at the REPL the same way a simulated reset always has.

machine.py's _flush_data() is deliberately not called here: it does settings.save(), and clear_seed() blanks settings on the line before fast_wipe(), so flushing would write back what was just erased.

Testing

• testing/test_sim_pyb.py — 7/7 pass
• Destroy Seed run end to end in the simulator, which now finishes and stops cleanly:

NOTE: One-way callgate into bootloader: method=23 arg2=48879
Simulator: would wipe seed & reboot. Exiting.

• Nothing under testing/ references oneway, fast_wipe, show_logout, or enter_dfu
• bare_metal.py installs its own my_oneway, so bare-metal mode is unaffected

Simulator-only — no changes to shared/ or any on-device code path.

[scgbckbone]

Thanks for the contribution. We already cover the Destroy Seed UX in testing/test_ux.py::test_destroy_seed. The destructive behavior itself needs to be verified on real hardware, so I don’t think emulating it in the simulator adds enough testing value to justify the additional code.