Use gpt-5.6-luna as the small-fix fallback

[coneilen]

Summary

The operator decided that if gpt-6-luna underperforms on small fixes, the fallback is gpt-5.6-luna.

Changes

• investigation/windows-parity-orchestration-plan.md
  • The small-fix fallback is now gpt-5.6-luna, high.
  • The small-fix escalation ladder is now gpt-6-luna, then gpt-5.6-luna, then gpt-6.1-sol, then claude-sonnet-5.5. The lane-implementation ladder is unchanged.

Test plan

RED: Select-String investigation\windows-parity-orchestration-plan.md -Pattern 'gpt-5.6-luna' on main 5c4a0c9 -> 0 matches; the small-fix fallback was gpt-6.1-sol
GREEN: same Select-String after the edit -> 2 matches (model table fallback and 1st escalation)
REGRESSION: git diff --check -> clean; git ls-files --eol -> CRLF preserved; bash Tools/ci/tests/classify-changes.test.sh -> 59 passed, 0 failed

Documentation-only change: no product code, so the Windows suites and macOS make test/make check were not run.

Checklist

• I have read the Contributing Guidelines
• I have signed off my commits (git commit -s) per the DCO
• Tests pass locally (make test)
• Code follows the existing style (make check)
• I added the test/contract before the implementation and observed the intended RED failure