[shin4141]

Summary

• Add --dry-run to jupyter nbsearch update-index.
• Reuse the existing notebook selection and filtering logic, then print selected relative notebook paths in deterministic order.
• Exit before notebook parsing or any Solr/S3 write.

Compatibility boundaries

• Dry-run still loads the existing Python config, matching normal update-index behavior.
• Existing basename-only matching for an explicit path is intentionally preserved.
• Selected paths are written to stdout; diagnostics remain on stderr.

Tests

• python -m unittest nbsearch.tests.test_update_index -v (8 tests passed)
• python -m unittest discover nbsearch.tests -v (22 tests passed)

Fixes #9

[github-actions]

👈 Launch a Binder on branch shin4141/nbsearch/codex/issue-9-update-index-dry-run