feat: make overnight-dent testable — --test-task, merge tracking, --experiment (kaizen #322)

[aviadr1]

Summary

• --test-task flag: synthetic fast task that creates a trivial PR instead of running /make-a-dent. Completes in <2 min for pipeline iteration.
• --experiment flag: extra diagnostics — main HEAD before/after pull, per-PR merge status tracking, auto-merge queue visibility.
• checkMergeStatus(): new exported function that checks PR state via gh CLI, returns merged/auto_queued/open/closed/unknown.
• buildPrompt() now exported and supports test_task mode.
• BatchState extended with test_task and experiment optional fields.
• 13 new tests (42 total, all passing).

Context

This is Step 0 of the overnight-dent merge pipeline investigation (PRD qwibitai#321). The overnight-dent batch runner creates PRs but doesn't merge them. Before fixing the pipeline, we need fast iteration tools to test hypotheses about why merges fail.

Test plan

• 42/42 vitest tests pass (29 existing + 13 new)
• 91/91 scripts/ tests pass (no regressions)
• TypeScript build clean
• buildPrompt with test_task=true produces synthetic prompt without /make-a-dent
• buildPrompt with test_task=false still uses /make-a-dent
• checkMergeStatus returns correct states for merged/queued/open/error cases
• --test-task and --experiment flags parsed in trampoline, passed to state.json
• Manual: ./scripts/overnight-dent.sh --test-task --max-runs 1 "probe" (deferred to Step 1)

scripts/overnight-dent.sh: bash script — tested indirectly via overnight-dent-run.test.ts (buildPrompt tests verify state.test_task handling)

Fixes Garsson-io/kaizen#322

🤖 Generated with Claude Code