chore(sdk): regenerate from the API — Plaid, disclosures, grouped search

[jfrench9]

Summary

Regenerated from the API. Everything lands in the generated tier and is additive except one relaxation, so this is a minor.

Changes

• Plaid: plaid joins the three provider enums, PlaidConnectionConfig arrives, and OAuthInitResponse gains link_token for providers that authorize in an embedded widget.
• auth_url relaxes from required str to nullable, because Plaid has none — the user authorizes in Link rather than at a redirect. It is the one change a consumer could feel; roboledger-app's call sites already guard on its absence.
• RoboLedger views: disclosures and information_block, with their request, response and envelope models.
• Search: group on the request, also_in_filings on a hit.
• Docstrings follow the API's own: connection limits name Plaid, and filed_at / filed_by now explain what a report shared in from another graph carries and what it cannot.
• robosystems_client/graphql/schema.graphql: docstring-only movement on those same two fields.

Breaking Changes

None to the stable tier. No facade, root export, error class, auth config, or emit-path symbol changed, so a consumer pinned >=1,<2 is unaffected. The auth_url nullability belongs in the release notes, since a strict consumer's type checker will see it.

Testing

• just test-all: ruff, ruff format, basedpyright all clean.
• just test: 591 passed, 17 skipped.
• just generate-graphql regenerated graphql/generated/ with no diff, so the CI drift gate has nothing to catch.

Certification

• I have the right to submit this work under the Apache 2.0 license, and do so. Where any part of it is owned by my employer, I have their permission.

🤖 Generated with Claude Code

https://claude.ai/code/session_011KBrhQHzwrB39C4c4Vw9Pq