iron tail: --sql, --streams, --schema, --json for agent introspection

[kualta]

Summary

Four new flags on `iron tail` so agents can introspect and query the OO hub without curl + config-fishing:

• `--streams` - list every stream with type and field count.
• `--schema STREAM` - print the schema (column name + type) of a stream.
• `--sql SQL` - run an arbitrary SQL query and emit results as NDJSON (one JSON object per line). `--since` still sets the time window; other filter flags ignored.
• `--json` - switch default tail output to NDJSON.

All four share `iron tail`'s existing hub-URL + credential plumbing.

Test plan

• cargo fmt, clippy -D warnings, test
• iron tail --streams: 5 streams listed with real field counts
• iron tail --schema flow_events: 8 columns + types
• iron tail --sql "SELECT service, count(*) FROM "_rumdata" ... GROUP BY service" --since 48h: NDJSON aggregation results
• iron tail --since 30s --json: raw JSON rows instead of colorized text