feat(ci): add typegen check workflow

[rikublock]

Summary

Workflow to check whether the generated frontend types in invokeai/frontend/web/src/services/api/schema.ts match those on the server.

Notes

There still appears to be some non-determinism in the typegen creation process. Depending on the Torch installation (CPU, GPU, etc.), the default value of an InputField may change.

Example: https://github.com/invoke-ai/InvokeAI/blob/main/invokeai/app/invocations/create_denoise_mask.py#L33

DEFAULT_PRECISION = TorchDevice.choose_torch_dtype()

fp32: bool = InputField(
    default=DEFAULT_PRECISION == torch.float32,
    description=FieldDescriptions.fp32,
    ui_order=4,
)

Merge Plan

• Address non-determinism (my suggestion: use a fixed default value) fix(app): fixed InputField default values #7464
• Ensure schema is up-to-date: chore(ui): update typegen schema #7462

Related Issues / Discussions

#7394 (comment)

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Tests added / updated (if applicable)
• Documentation added / updated (if applicable)
• Updated What's New copy (if doing a release after this PR)

[psychedelicious]

Awesome, thank you.

And perfect timing! The OpenAPI schema changed with the new FLUX ControlLoRA models, but the TS types weren't updated. In #7462 once I rebased, there were typescript errors. I fixed the issue in the frontend code.

[psychedelicious]

Doing some tests in #7467 - please do not merge this just yet.

[psychedelicious]

Whoops, I accidentally experimented on your branch, fixed that. Tested - looks good. Thanks!

How I tested:

• Add a new node without regenerating the TS types.
• Confirm the new check fails.
• Regenerate the TS types.
• Confirm the new check passes