Tags: tw93/Mole

release: prepare V1.69.0

Bump VERSION and the security audit date for V1.69.0, and record the
BytesSI rounding fix (#1914, shipped in V1.68.0) as incident I-32 with
TestBytesSI as its catcher, as the pre-release review asked.

Verified: check.sh --format, full Bats suite (EXIT=0, no skips), go test,
make build, go1.26.10 release builds with check_release_minos.sh (12.0 on
both architectures) and scripts/e2e_macos.py 10/10 on this Mac.

Rollback: delete the V1.69.0 release and tag, then revert this commit.

release: prepare V1.42.0-windows

Package the work since V1.41.0-windows: mo optimize reads SFC's verdict
from CBS.log when sfc.exe prints a localized summary, so a clean scan on
zh-CN Windows is no longer logged FAILED (#1915).

Verified: both version sources read 1.42.0; Check and Validation passed on
7097023 with Pester on both Windows engines, and Optimize.Tests passed
58/58 in the ARM64 VM under Windows PowerShell 5.1 and pwsh 7 with a real
sfc /scannow read back from CBS.log. The full suite on the ThinkPad,
native e2e and the signed package are checked before the tag is published.

Rollback: revert this commit before tagging; after tagging, delete the
draft release and V1.42.0-windows tag, then revert.

docs(incidents): record the update that kept stale modules behind a s…

…uccess message

I-29 names the #1905 squash commit e50a170, which the contributor
branch's own row could not reference before the merge; I-28 was taken
on main by the Touch ID dialog fix in the meantime.

Verified: scripts/audit_incidents.py passes against this checkout.

chore: update contributors [skip ci]

V0.3.0-linux Tidy

Mole CLI (Linux) 0.2.0

Mole CLI (Linux) 0.1.0

docs(incidents): record the chrome, brave and adobe caches cleaned in…

… use

Both rows cleaned caches a running app had open in published releases;
the fixes are 12f10aa7 and aa734f19. Chrome and Brave should have been
caught at write time, since the Application Support rows beside them
already used the process guard; the Adobe row had no guard at all and
belongs to the destructive-sink review. The Yarn row change gets no row:
the old path was always empty, so it removed nothing.

Verified: scripts/audit_incidents.py and scripts/check.sh --no-format
pass.

test(e2e): read the fixture release through gh when anonymous calls a…

…re limited

The anonymous GitHub API allows 60 calls an hour per egress IP, and the
shared proxy had used them up, so the release run stopped at the fixture
download before it reached the PC. A signed-in gh has its own quota and
is used only after a 403 or 429; the asset still downloads from its
public URL and is checked against the published digest as before.

Verified: tests/e2e_harness_test.py passes, and its new case fails on
the old script.