chore: update catalog.yaml with recent fixes

[delebedev]

Summary

Update catalog.yaml to reflect PRs #214-#221:

• cleanup-discard: partial → wired (3 tests re-enabled)
• surveil: multi-card surveil 2+ fix noted
• stack-resolution: game-over detection after resolution
• declare-blockers: zero-blocker auto-advance
• discard: SelectNReq fields fixed, partial → wired
• mill: affectorId populated from stack source

🤖 Generated with Claude Code

[github-actions]

Test Results

  227 files  ±0  227 suites  ±0   2m 23s ⏱️ ±0s
  995 tests ±0  963 ✅ ±0   32 💤 ±0  0 ❌ ±0 
1 776 runs  ±0  977 ✅ ±0  799 💤 ±0  0 ❌ ±0 

Results for commit 8545f99. ± Comparison against base commit c58f0b2.

[github-actions]

CI Report — Gate

Tests: 815/815 passed (189 skipped)

Coverage: 10.5% (615/5852 lines)

Module	Coverage	Lines
tooling	🔴 11%	615/5852

Slow tests (>3s): 1

• InstanceIdReallocTest.Limbo grows across multiple plays (5.5s)

[delebedev]

@claude please review this ralph PR

[claude]

Claude finished @delebedev's task in 1m 8s —— View job

---

Review: chore/catalog-update

• Gather context
• Read changed files
• Review catalog.yaml changes
• Post review feedback

---

Clean catalog maintenance PR. All green (815/815 tests). Two files changed.

.gitignore (+1 line)

Adds .claude/worktrees/ — correct, these are ephemeral agent worktrees that shouldn't be tracked. No issues.

docs/catalog.yaml — 7 mechanic updates

All status promotions and note additions look accurate against the PR summary. Going through each:

cleanup-discard: partial → wired
Good. The old note flagged a stall bug. New note correctly records the fix and cites the 3 re-enabled tests (PR #214). Promotion is backed by concrete test evidence.

surveil (status unchanged, notes extended)
The added detail about game.findById vs player.allCards for zoneless cards is useful technical context. Good.

stack-resolution (status unchanged, notes extended)
One minor readability nit: "PR #221, was #122" — it's slightly ambiguous whether #122 is the original issue or an earlier PR. Consider "(PR #221; fixes issue #122)" if clarity matters. Not blocking.

declare-blockers (status unchanged, notes extended)
Zero-blocker auto-advance is a real behavioral edge case. The note accurately describes the behavior. Good.

discard: partial → wired
The old note explicitly flagged SelectNReq field mismatches as a known bug. Removing it and promoting to wired is appropriate given PR #215 fixed the wire fields. The plan file reference (docs/plans/2026-03-21-discard-wire-spec.md) is now dead in the notes — which is fine since the plan was executed. The file itself presumably still exists but no longer needs a catalog pointer.

mill (status unchanged, notes extended)
affectorId from stack source — correct, concise. Good.

Gaps noticed

PRs #216 and #220 from the stated range (#214–#221) aren't reflected in the catalog. This may be intentional (tooling or infra PRs with no catalog-visible mechanic impact), but worth a quick sanity check that nothing was missed.

Summary

LGTM. The catalog accurately reflects the shipped fixes, status promotions are evidence-backed (tests re-enabled), and CI is clean. The one optional nit on the was #122 phrasing in stack-resolution is purely cosmetic.