fix(nats): derive CloudEvents subject from metadata instead of hardcoded TODO

[TRIVENI206]

…ded TODO

Summary

This PR replaces the hardcoded CloudEvents subject ("TODO") in the NATS eventing driver with a dynamically derived value from message metadata.

The subject is now determined using the following priority:

• entity_id
• repository_id or artifact_id
• project_id

If none of these values are present, a fallback value ("minder") is used.

This improves event filtering in NATS JetStream and aligns with the CloudEvents specification, while keeping the change minimal and avoiding unrelated refactoring for easier review.

Fixes #6250

Testing

• Built the NATS module using:
  go build ./internal/events/nats/...

• Ran tests for the NATS module:
  go test ./internal/events/nats/...

• Verified that:

  • The code compiles successfully
  • No tests in the NATS module are broken
  • The subject is now set dynamically instead of being hardcoded

[CLAassistant]

All committers have signed the CLA.

[evankanderson]

This seems to be conflicting with the work discussed in #6251 (and runs into the same question as to whether we can get by with simply entity_id given our simplification and standardization on entities last year.

[coveralls]

coverage: 58.318% (+0.003%) from 58.315%
when pulling ed656cf on TRIVENI206:fix-nats-subject
into 0cbed8a on mindersec:main.

[TRIVENI206]

Thanks for the feedback!

I based the implementation on multiple metadata fields for broader compatibility, but I see the point about standardizing on entity_id.

Would you prefer simplifying this to use only entity_id (with a fallback if missing), or keeping the current priority-based approach?

Happy to update the PR accordingly.