CLI: add `hflow stale --exit-code` so CI can gate on stale episodes

[kstonekuan]

Current behavior

_command_stale (src/hflow/cli.py:268) prints stale episode URIs and always returns 0 (src/hflow/cli.py:318), regardless of how many episodes are behind the current pipeline version. That means hflow stale cannot gate a CI job or a pre-training checklist, while hflow doctor already exits 1 on findings.

Proposed change

Add an --exit-code flag (named after git diff --exit-code): when passed, exit 1 if any stale episode was printed. Keep the default at 0 so existing pipes keep working (docs/CATALOG.md documents piping the output into xargs hflow ingest).

Definition of done

• hflow stale --exit-code ... exits 1 when at least one stale episode exists and 0 when none do.
• Without the flag, behavior is unchanged.
• Test with the existing catalog fixtures in tests/test_catalog_curation.py.

Validation

uv run pytest -q
uv run ruff check --fix && uv run ruff format
uv run ty check

[chiruu12]

Happy to take this one. Plan: add --exit-code (store_true) to the stale subparser and return 1 from _command_stale when it's set and at least one stale episode was printed. Default stays 0 so the xargs hflow ingest pipe in docs/CATALOG.md is unaffected. I'll cover the behind and up-to-date cases against the existing fixtures in test_catalog_curation.py and document the flag.

[kstonekuan]

Yours, @chiruu12, assigned. The plan: store_true on the stale subparser, return 1 from _command_stale only when the flag is set and at least one stale episode printed, default exit code unchanged so the xargs pipe in docs/CATALOG.md keeps working.

[kstonekuan]

Heads-up @chiruu12: #30 is still yours, but someone else opened a pull request for it (#76, now closed since you have the assignment). No pressure and no deadline, I just wanted you to know there is interest, in case you would rather hand it off than hold it.

If you have hit a snag, say so here and I will help. If you would rather do something else, that is completely fine too, just let me know and I will unassign so it goes back in the pool.

[chiruu12]

Thanks @kstonekuan, and thanks for handling #76 the way you did.

No snag, I want to keep it. The reason there was no PR yet is that I wanted the change verified rather than just written. The flag itself is a few lines, but the part worth getting right is that the default path stays exactly what it was, since docs/CATALOG.md pipes stale into xargs hflow ingest and a changed exit code there breaks that quietly. So I checked the new tests fail on current main with unrecognized arguments: --exit-code before checking they pass on the change, and I ran the whole suite on main as well as on the branch so the comparison is against a measured base rather than an assumed one.

PR is up now: #81

Local run on macOS with uv sync --locked --all-extras:

• uv run pytest -q tests/test_catalog_curation.py -k cli_stale, 4 passed
• uv run pytest -q, 301 passed, 5 failed, 6 skipped. The same 5 fail on unmodified main on this host, all ffmpeg related, and base sits at 299 passed, so the change adds two and moves nothing else
• uv run ruff check and uv run ruff format --check, clean
• uv run ty check, clean