Take the OWID with the dated key fetch, and report whether a licence key was given

[jwrosewell]

Summary

Moves the owid-python submodule onto the fork head carrying the dated public key fetch and the key schedule, and adds the one client accessor this package was missing against Java and .NET.

Draft while the specification alignment review runs, because that review may add commits here. See Still to come at the end.

What changed

• owid-python submodule 0a77e39e to 5fe9a7cc, being the merge of Add the dated public key fetch and the key schedule Python had none of SWAN-community/owid-python#4, which gave the Python port the dated fetch and the key schedule the other ports already had.
• has_licence_key on DidClient. Java has hasLicenceKey() and .NET has the same, and this package had nothing. A licence key is needed only by redeem, so a client without one still reads keys and verifies signatures, and a caller can now ask rather than track it alongside.

What deliberately did not change

DidClient does not call the new PublicKeySchedule. It fetches the whole published schedule from the cloud's own key endpoint with a resource key and selects by the start of a key's period itself, applying a fifteen minute boundary tolerance that the OWID library does not model. The two selection rules are different on purpose and both are correct, and the Java and .NET clients keep their own selection for the same reason.

The build time copy into the package as the private module _owid takes whatever Python files the submodule holds, so public_key_fetch.py and public_key_schedule.py are now carried in a published wheel with the rest. ci/copy-owid-source.ps1 needed no change.

Verification

python -m pytest tests in fiftyone_pipeline_did with the submodule at 5fe9a7cc: 145 passed, 2 skipped.

Still to come in this pull request

The specification alignment review across SWAN-community and 51Degrees, checking that 51Did is properly defined and that the implementations agree. Anything it finds for this package lands here rather than in a separate pull request.

One finding is already recorded and is not fixed here. The five outcomes of a detailed offline signature check are spelled three different ways across .NET, Java and Python, which does not meet the convention the packages state themselves, that the backing string is the cross language name of the outcome so it can be logged or carried between services. Aligning them is a breaking change in Java and Python and is a decision rather than a tidy up.

Produced with AI assistance and needs human review before merging.

[jwrosewell]

Correction to this pull request's description and commit message. The first paragraph says this change moves the owid-python submodule to 5fe9a7cc. It did not. Automation51D's submodule update #73, merged at 02:35 UTC on 6 September 2026, had already moved main to that commit, so by the time #74 merged its branch carried no submodule change and the squash commit 30f94c0 touches only did_client.py. The description of what 5fe9a7cc contains is accurate, it just landed through #73. Found in review after merge.