feat: add /flight-review, /flight-compare, /flight-annotate slash commands

[lewisnsmith]

Summary

• Adds three new Claude Code slash commands to src/setup.ts SLASH_COMMANDS array
• /flight-review: structured session critique — retries, errors, tool overuse, good decisions
• /flight-compare: diffs two experiments with a 3-bullet summary (winner, biggest delta, suggested next test)
• /flight-annotate: labels each turn and emits flight annotate shell commands to persist labels (strict one-command-per-turn prompt to mitigate R6 deviation risk)
• Also updates the two existing slash command bodies (flight.md, flight-log.md) to reference flight logs (not flight log) ahead of the PR 1 rename
• Makes the remove status line generic: "Removed Flight slash commands"

Dependencies

Depends on feat/cli-logs-rename (PR 1) and feat/experiment-registry (PR 2). Land this last.

The slash command bodies reference flight logs verbose, flight show, and flight experiment diff — all of which land in PR 1 and PR 2 respectively.

Test plan

• npm run typecheck passes
• npm run lint passes
• test/setup.test.ts — 7 tests pass (3 original + 4 new)
  • After runSetup with slashCommands: true, all 5 .md files exist
  • No body references flight log (singular with space)
  • Bodies reference correct CLI verbs (flight logs audit, flight logs verbose, flight show, flight experiment diff, flight annotate)
  • After runRemove, all 5 files are deleted
• After flight claude setup, all 5 slash command files exist in ~/.claude/commands/
• After flight claude hooks remove, all 5 are deleted

[lewisnsmith]

@copilot resolve the merge conflicts in this pull request

[lewisnsmith]

@copilot resolve the merge conflicts in this pull request

[Copilot]

@copilot resolve the merge conflicts in this pull request

Resolved and pushed in 3ab0d9c. I merged origin/main into this branch, fixed the conflicts in CHANGELOG.md and CLAUDE.md, and verified everything passes with npm run check.