feat: Add is_producer_url_unstable to feed schema and expose in API

[jcpitre]

Closes #1766

• Added is_producer_url_unstable to the DB schema (nullable boolean, default null).
• Modified the populate scripts to accept the new flag.
• Exposed the flag in the public API, in the /v1/feeds, /v1/feeds/{id} and /v1/search responses (via source_info).
• Also exposed/handled it in the internal Operations API (source_info + update-request models).

The rule for updating the DB from the CSV file is:

• An empty is_producer_url_unstable cell in the CSV is a no-change in the DB.
• An explicit TRUE or FALSE value in the CSV overwrites the current value.
• This matches the existing behavior of is_official (get_safe_boolean_value + a not-None guard).

Note: This PR should be merged before this PR from mobility-database-catalogs: Add unstable_producer_url field
#1547

Testing tips:

This PR includes new unit tests.
The whole chain (mobility-database-catalogs → mobility-feed-api) was tested somewhat, except the notification between the repos, which is outside of scope.
The test was:

• In a branch in mobility-database-catalogs, add the flag to the JSON files:
  • "is_producer_url_unstable": "True", to one JSON file, and set it to "False" in another.
• Run the Export catalogs to CSV workflow.
  • Note this workflow was changed so it can accept a file name other than sources.csv. The file is placed in the mdb-csv bucket (authentication required).
• Run the Mobility Catalog Update workflow in this repo.
  • Note this workflow was modified to accept the file name that was used when creating the CSV.

Tested these cases:

Catalog JSON value	Existing DB value	Resulting DB value	Note
True	any	true	explicit value always wins
False	any	false	explicit value always wins
absent / empty	null	null	new/unknown feed stays unknown
absent / empty	true	true	retained — empty does not clobber
absent / empty	false	false	retained — empty does not clobber

Please make sure these boxes are checked before submitting your pull request - thanks!

• Run the unit tests with ./scripts/api-tests.sh to make sure you didn't break anything
• Add or update any needed documentation to the repo
• Format the title like "feat: [new feature short description]". Title must follow the Conventional Commit Specification(https://www.conventionalcommits.org/en/v1.0.0/).
• Linked all relevant issues
• Include screenshot(s) showing how this pull request works and fixes the issue(s)

[davidgamez]

[nitpick]: Taking into consideration that None is a valid value and has a "meaning", this conditional doesn't allow setting the value back to None.

[jcpitre]

After discussion with @emmambd, we don't need to go back to None.

For clarity (and my own understanding), here is a trace of the value through each stage.

1. Export — JSON → CSV (export_to_csv.py)

JSON value	CSV cell
true	True
false	False
null	(empty)
field absent	(empty)

An explicit null and a missing field are indistinguishable after export — both become an empty cell. The CSV never contains the literal string null.

2. Populate — CSV → DB (populate_db_gtfs.py)

• DB default: NULL
• Empty cell → no-op — the DB value is left unchanged (so NULL stays NULL, and an existing True/False is retained).
• True / False → written to the DB (overwrites whatever was there).

3. Read — DB → API response

DB	Python	API JSON
NULL	None	null
true	True	true
false	False	false

Net behavior

A null or absent value in the JSON results in:

• true / false if the DB value was already true/false, or
• null if true/false was never set.

You can't go back to the null API response once it's set to True or False.