fix(session): investigate hanging review and plans commands in large repos

[4444J99]

Failure state

In organvm-iii-ergon/public-record-data-scrapper, both required session-close commands hang with no output and require an external timeout:

• organvm session review --latest
• organvm session plans --project .

Observed on 2026-04-02 from repo root. Both exit 124 when wrapped with timeout 20, producing no stdout/stderr before termination.

Why this matters

These commands are mandated by repo session-close protocol in CLAUDE.md. A silent hang means session closeout cannot be completed deterministically, which creates governance and memory-loss risk.

Reproduction

cd ~/Workspace/organvm-iii-ergon/public-record-data-scrapper
timeout 20 organvm session review --latest
timeout 20 organvm session plans --project .

Expected

Commands should either complete or fail with actionable diagnostics.

Actual

Both commands block until killed by timeout.

Related context

• organvm prompts distill --dry-run is not hanging, but fails fast because data/atoms/clipboard-prompts.json is missing and tells the operator to run organvm prompts clipboard first.
• This was discovered during closeout of IRF-APP-003 work for public-record-data-scrapper.

Suggested investigation

• Reproduce under profiler / debug logging.
• Determine whether the stall is transcript discovery, plan indexing, git inspection, or corpus path resolution.
• Add timeout / progress diagnostics so the failure mode is visible even before root cause is fixed.

[4444J99]

Backlog engagement 2026-07-19 — disposition: evolving.

Verified backlog state: the active fix branches (#111 and #132) both target session/plans.py, git/status.py, and session review/plan tests, but they are conflicting/dirty and red. I did not verify the large-repo hang is fixed on current main; the next technical step is a fresh rebase plus a bounded regression test that runs the review/plans commands under timeout against a representative large repo fixture.