fix: align temporary chat tools files and RAG behavior

[chenyme]

Summary

Closes #687.

Align temporary chat behavior with normal conversations while preserving its non-persistent privacy boundary.

• Enable model-native tools and user-configured model parameters in temporary chat.
• Enable selected MCP tools with request-scoped tool execution.
• Enable Skills and ephemeral knowledge base RAG retrieval.
• Support temporary image and document attachments without creating persistent file records.
• Extract temporary documents for full-context injection and send images as multimodal input.
• Support editing and retrying temporary messages locally without creating conversation branches.
• Keep temporary feedback local instead of writing message feedback records.
• Reuse the normal conversation error diagnostics including localized summaries and sanitized upstream request and response details.
• Fix knowledge base RAG being incorrectly disabled together with temporary attachment RAG.
• Preserve billing and required audit records without persisting conversations messages runs or uploaded files.
• Clean up temporary files readers previews and generated in-memory state after completion cancellation or page exit.

Change type

• Bug fix
• Feature
• Documentation
• Refactor
• Configuration / deployment
• Security hardening
• Other

Affected areas

• Frontend / UI
• Backend / API
• Authentication / authorization
• Conversations / streaming
• Files / RAG / extraction
• Model routing / providers
• MCP / tools
• Billing / payments
• Admin console
• Deployment / Docker / configuration
• Documentation

Verification

• cd backend && go test ./...
• cd frontend && pnpm typecheck
• cd frontend && pnpm lint
• cd frontend && pnpm build

Screenshots, API examples, or logs

Temporary chat now supports the same configured model capabilities MCP tools Skills and knowledge bases as normal chat.

Upstream failures now render the standard generation error card with an HTTP summary and expandable sanitized request and response details.

Configuration, migration, and compatibility notes

• No database migration is required.
• No new configuration is required.
• The temporary chat endpoint remains compatible with existing JSON requests.
• Multipart requests are now accepted for request-scoped temporary attachments.
• Temporary attachments are not written to file metadata object storage or background processing queues.
• Temporary documents use full-context extraction rather than persistent attachment vector indexing.
• Selected knowledge bases continue to use the configured RAG and embedding services with ephemeral retrieval.
• Provider state cache identifiers and persistent response state remain disabled for temporary requests.
• Swagger and generated API contract artifacts were updated.

Documentation

• Documentation is not needed for this change.
• Documentation was updated.
• Documentation still needs to be updated.

Security and privacy

• No secrets, tokens, credentials, local config, or personal data are included.
• User data access remains scoped by authenticated user context unless an admin-only path explicitly requires broader access.
• Security-sensitive behavior was reviewed, including authentication, authorization, provider routing, file processing, billing, and admin APIs where relevant.

Temporary conversations still do not persist conversation messages runs provider response state or request-scoped attachments. Necessary billing and audit records remain intact.

Checklist

• I searched existing issues and pull requests.
• Changes are focused and do not include unrelated refactors.
• Tests or static verification were run where practical.
• User-facing behavior, deployment steps, API contracts, or configuration changes are documented.
• Generated artifacts are included only when this project explicitly requires them.
• Caches, build output, .pyc files, .env files, and local storage data are not committed.