Convert predictions to cliplabels.json using movement

[sfmig]

Description

What is this PR

• Bug fix
• Addition of a new feature
• Other

Why is this PR needed?
To convert prediction files supported by movement to cliplabels.json files, that we can clip later when extracting video clips.

What does this PR do?
It adds a predictions_to_poseinterface function that reads prediction files as movement datasets and exports the trajectories as a cliplabels.json file, defined for the full range of frames in the prediction file.

References

\

How has this PR been tested?

Tests pass locally and in CI.

Is this a breaking change?

No.

Does this PR require an update to the documentation?

No.

Checklist:

• The code has been tested locally
• Tests have been added to cover all new functionality
• The documentation has been updated to reflect any changes
• The code has been formatted with pre-commit

[lochhh]

Thanks @sfmig , I think the decision to use movement for loading predictions and sleap-io for loading annotations makes a lot of sense! Most suggestions are trivial, but the main comments are related to the lack of error-handling in predictions_to_poseinterface and tests that exercise the function's various possible behaviours.

[niksirbi]

FYI I've now merged neuroinformatics-unit/movement#920 into movement, which adds a public infer_source_software function, and a source_software='auto' option to load_dataset.

Once we release, this PR can be rebased to directly use movement for the 'guessing'.

[niksirbi]

@sfmig when you get the chance to get back to this, you may find it useful to cherry-pick the following commits from PR #49:

• c104ec2 (#49) (simplify predictions_to_poseinterface using movement>=0.16 auto source inference)
• e9aa17e (#49) (Rename cliplabels to videolabels in predictions_to_poseinterface)

[sfmig]

Thanks @lochhh and apologies for the delay. I think all comments are addressed, will re-request now.

[lochhh]

Thanks again @sfmig for addressing all comments! Pre-approving as there are only two minor things: duplicate movement dependency in pyproject.toml, suggestion to remove assert function calls (wiring will be tested in integration tests).