feat(transport): add CEP-15 i/k discovery tags

[abhayguptas]

Summary

This PR is the next stacked CEP-15 step on top of #38.

It adds the NIP-73 discovery tags described in CEP-15 to public kind:11317 tools/list announcements. For opted-in common-schema tools, the SDK now publishes:

• ["i", schemaHash, toolName] for each implemented common schema
• ["k", "io.contextvm/common-schema"] once per event

This complements the _meta.schemaHash work from #38 by making common-schema providers discoverable directly through announcement tags.

Changes

• extend withCommonToolSchemas() with a pure CEP-15 i/k tag producer
• add a generic list-tools announcement tag seam in NostrServerTransport
• attach those tags to public tools/list announcement events in AnnouncementManager
• add unit tests for i/k tag generation
• extend the transport integration test to verify the outer announcement tags

Tests

• bun test src/transport/server-transport-common-schemas.test.ts src/transport/nostr-server-transport.test.ts
• bun run typecheck

Notes

This PR is intentionally scoped to outer announcement discovery tags only.

It builds on:

• feat(core): add CEP-15 common schema helpers #37 for schema normalization + hashing helpers
• feat(transport): inject CEP-15 schemaHash into tools/list results #38 for _meta.schemaHash injection and announcement-content consistency

[abhayguptas]

cc - @ContextVM-org

[ContextVM-org]

The main improvement I would suggest is in server-transport-common-schemas.ts: the schema hash is currently recomputed in both createCommonSchemaToolsResultTransformer() and createCommonSchemaAnnouncementTagsProducer(). It is not a big issue, but it does feel like a bit of duplicated work and duplicated logic inside the same helper module, so I would consider extracting that into one small shared internal helper. I also think it would be nice to make sure the no‑op path in the transformer stays cheap, since right now it looks like it may still allocate metadata objects even when the schema hash is already present.

[abhayguptas]

Thanks, agreed. I extracted the schema-hash access into a small shared internal helper in server-transport-common-schemas.ts and updated the announcement tag producer to reuse that path instead of duplicating the inline hash logic.

I also kept the transformer no-op path cheap and added focused coverage for both the already-present schemaHash case and the announcement-tag reuse path.

[ContextVM-org]

How would you like to proceed with your prs? Should we just merge one by one or merge them all in #44 and merge them all together?

[abhayguptas]

We can merge them one by one in stack order. #44 contains the earlier commits in its history, but I think the separate PRs are easier to review and reason about since each one stays focused on a single step.

[ContextVM-org]

For this time is ok, but I think that for scoped prs like this one is better to keep them in a single pr, because currently there are some comments that are sparsed between different prs and makes this a bit awkward to track

[abhayguptas]

That makes sense, agreed. For this round we can keep the scoped CEP-15 changes consolidated here so the review stays easier to follow, and from next time I’ll avoid splitting work like this across multiple PRs unless it is really necessary.

[ContextVM-org]

Last thing before merge this pr, add a changeset

[abhayguptas]

@ContextVM-org hi, I have added the changeset. Please review it and if everything looks good then we can merge those PR's.

[ContextVM-org]

Just some minor comments before merge.

[abhayguptas]

Just some minor comments before merge.

fixed, please review