Preserve and recover archive frame IDs for calibrations

[timbeccue]

Problem

Archived calibrations can have calimages.frameid = NULL because two registration commands omit the archive ID when constructing their DataProduct:

• Importing BPMs with banzai_populate_bpms --db-address "$DB_ADDRESS".
• Uploading a manually prepared calibration with banzai_add_super_calibration /path/to/readnoise.fits --db-address "$DB_ADDRESS" --upload-to-archive.
• Re-importing an archived calibration can also overwrite an existing ID with NULL.

A reduction can find these calibrations through the Archive basename lookup, but the existing save-back condition never persists the recovered ID: database records already include a frameid key, and opening the file updates that dictionary in place.

Changes

• Pass the Archive response's id into DataProduct in add_bpms_from_archive.
• Pass the ingester response's frameid into DataProduct in add_super_calibration, removing the ineffective cal_image.frameid assignment.
• Remember whether a master calibration's ID was missing before opening it, then use the existing database helper to save a successfully recovered ID.

Recovery writes to cal_db_address, the same database used to select the calibration. This defaults to db_address in the main AWS configuration; an explicit calibration database remains the target at sites.

Registration without upload still stores NULL, including when replacing an existing filename. Existing NULL rows can be repaired through corrected re-imports or basename fallback during calibration use.

[cmccully]

Can we add if frame_id_was_missing case that then populates master_calibration_image.frame_id in the database?

[timbeccue]

The db update is just below in dbs.update_calibration_frameid(...). The old condition prevented it from running because it just looked for whether the key existed and not whether it was populated vs None, which is what this change fixes.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Add coverage for re-importing an existing filename and updating its frame ID instead of NULL.

Review effort: Lite
Findings: None

What changed in this PR

This PR preserves archive frame IDs during calibration registration and recovery.

Changes:

• Passes archive and ingester IDs into calibration records.
• Persists IDs recovered through archive lookup.
• Adds database tests for registration and recovery.

File	Description
banzai/​tests/​test_dbs.py	Tests registration and recovery behavior.
banzai/​main.py	Preserves frame IDs during calibration registration.
banzai/​calibrations.py	Saves IDs recovered through archive lookup.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[cmccully]

I think this looks fine. I do want to make sure that we store the newly found frame id in the database and I'm not finding the code to do that. We should probably also add a change log note and bump the version (though we can wait to tag the final version until we've merged all the prs).