feat(ingest): guide users from dataset to creating a use case

[LukasWodka]

What

Rewrites the CLI's end-of-ingest "What's next" message so it bridges the mental-model gap that confused a pilot customer ([redacted]/[redacted]): ingesting registers a standalone dataset — to train or benchmark on it you must create a use case and select that dataset. The old copy ("ready for training jobs", metadata link only) implied the data would just become trainable, which it doesn't.

Before (internal/submit/summary.go):

What's next
View it in the dashboard: https://ai.tracebloc.io/metadata
Your dataset is ready for training jobs.

After:

What's next
Your data is registered as a dataset. View it at https://ai.tracebloc.io/metadata
To train or benchmark models on it, create a use case at https://ai.tracebloc.io/my-use-cases
Pick this dataset when you set it up.

Notes

• Prod-hardcoded on purpose — dev/staging are internal-engineer-only (they know the URLs); no env-plumbing added (decision: @LukasWodka).
• URL is /my-use-cases (the real Next.js route; /myusecases 404s).
• Copy-only in summary.go; go build + go test ./internal/submit/... green; gofmt clean.

Part of the onboarding-discoverability fix in backend#1073 (item 1). The backend error-message + docs items in that ticket are separate follow-ups.

🤖 Generated with Claude Code

---

Note

Low Risk
User-facing copy only in the ingest summary printer; no logic, URLs beyond hardcoded prod links, or data-path changes.

Overview
Updates the CLI "What's next" block after a successful ingest so it matches the product flow: ingest only registers a dataset, not a trainable job.

Instead of implying the data is ready for training jobs with a single metadata link, it now states that registration explicitly, links /metadata to view the dataset, and directs users to /my-use-cases to create a use case and select that dataset for training or benchmarking. A hint reminds them to pick the dataset during setup.

^{Reviewed by Cursor Bugbot for commit 63ec886. Bugbot is set up for automated code reviews on this repo. Configure here.}