feat: osa ingestion status / list — observe a run after starting it

[rorybyrne]

osa ingestion has only a start subcommand. Once a run is started there is no supported way to find out what it did.

Environment: osa-py 0.8.0.

Current behaviour

$ osa ingestion start --convention global-seismic-events --limit 25
 ✓ Starting ingestion for global-seismic-events  urn:osa:localhost:ing:4df450f2-...  0.1s

$ osa ingestion --help
╭─ Commands ──────────────────────────────────────╮
│ start  Start an ingestion run for a convention. │
╰─────────────────────────────────────────────────╯

It prints a URN and exits. There's no status, no list, and the URN can't be passed to anything.

What checking a run actually took

docker logs my-archive-server-1 2>&1 | grep -E "4df450f2"

and then, because the run's own COMPLETE line reported a count that turned out to be wrong (linked below), querying Postgres directly to establish the truth:

select count(*) from records;
select count(*) from metadata.seismic_event_v2;
select count(*) from features.estimate_energy;

That's three levels below the CLI to answer "did my ingestion work". osa logs server helps, but it means reading interleaved request logs and knowing which line matters.

Suggested fix

• osa ingestion status <urn> — stage, pulled / passed / published / failed counts, and any batch error.
• osa ingestion list — recent runs per convention with their outcome.

The data is already there: ingest_runs is a table, the workflow logs per-stage transitions, and process_batch already computes the counts. This is mostly surfacing.

Related: the counts these commands report should mean queryable rows, per the linked reporting bug — otherwise status inherits the same problem the log line has.

[rorybyrne]

Split out of the first-run report in #195, which has the full context and an index of the related issues.