[DRAFT] Handle None preds when no target channels for a stream

[shmh40]

Illustrative hack to fix current problem of generating None preds with no target channels for e.g. synop physical prediction for finetuned JEPA trained model case. Not sure that this works completely, currently writing out zarrs with 0s at evaluation.

Description

Some illustrative code to highlight the issue of None preds in some fine-tuning of JEPA models cases.

Issue Number

Closes #1724

Is this PR a draft? Mark it as draft.

Checklist before asking for review

• I have performed a self-review of my code
• My changes comply with basic sanity checks:
  • I have fixed formatting issues with ./scripts/actions.sh lint
  • I have run unit tests with ./scripts/actions.sh unit-test
  • I have documented my code and I have updated the docstrings.
  • I have added unit tests, if relevant
• I have tried my changes with data and code:
  • I have run the integration tests with ./scripts/actions.sh integration-test
  • (bigger changes) I have run a full training and I have written in the comment the run_id(s): launch-slurm.py --time 60
  • (bigger changes and experiments) I have shared a hegdedoc in the github issue with all the configurations and runs for this experiments
• I have informed and aligned with people impacted by my change:
  • for config changes: the MatterMost channels and/or a design doc
  • for changes of dependencies: the MatterMost software development channel

[shmh40]

@grassesi do you have any ideas on this and issue #1724? The problem is described in the issue, let me know if you have any questions. This PR is not intended to be merged, just highlights where the problem is in the code and how to immediately hack around it.