Make the stored-id prepass report what it did

[ajslater]

Follow-up to #840, which fixed neither symptom in practice. Both of its fixes hooked comicbox's event-emitting search session, and a re-tag never enters that session: comics codex already holds an issue id for are fetched by explicit id in the stored-id prepass, outside it. On a re-tag that prepass is the entire run.

Evidence from the reporting instance: the frozen snapshot had five comics, all {"comicvine": "matched"} with no Metron cell, merge_all_sources: true — while the log for the same run showed online metron: added id=134167, id=170212, id=8862… Metron had refreshed every one of those comics.

A merged refresh now credits every source that landed

record_prefetch_match took a single primary source. Under merge-all the prepass pins every stored id and comicbox fetches and merges them all, so a comic refreshed from both sources showed Matched in one column and an em-dash in the other — the cell that reads "never consulted".

resolved_id_sources reports which of the requested ids came back in the merged record, and every one of them is credited. The primary needs no such check: fetch_tags_by_explicit_id already returns nothing unless its own id landed. A source this scan never pinned — first-wins, or no credentials — still reports nothing, which is what the em-dash is for.

This matches what the search path already does: under merge-all comicbox emits one AutoWritten per contributing source, and the fold accumulates them.

The status rail covers the prepass

Only collect_results opened an OnlineLookupStatus, so a scan that resolved everything from stored ids ran start to finish with no active row at all — an empty rail, while the Tagging tab's table, published from the same loop, showed each lookup as it happened. #840's looking up on <source> subtitle was written to self._pass_runner.lookup_status, which is None for the whole prepass.

The prepass now opens the row (begin_status), totalled over the whole batch so the count doesn't jump when the search pass takes over, and advances complete per committed comic. collect_results adopts an already-open row instead of starting a second one, so the rail shows one continuous job whose elapsed time counts the prepass, and still finishes it in its finally. finish_status() in run_session's finally closes the row if the prepass raises before the search pass can adopt it.

Testing

• pytest: 1113 passed. New coverage: merged-fetch attribution, first-wins attribution (the source that sat out stays blank), the prepass opening and naming its status row, collect_results adopting rather than restarting it, and finish_status idempotence.
• make fix, make lint, make ty: clean.

🤖 Generated with Claude Code