feat: add success only duration billing for video generation

[chenyme]

Summary

Add duration-based billing for video generation while ensuring only successfully delivered videos are charged.

This change:

• Restricts duration pricing to active models with the video_gen capability.
• Validates duration pricing in both the admin UI and backend.
• Uses provider-reported video duration when available.
• Falls back to the requested duration, with 6 seconds as the default.
• Rounds fractional durations up to the next whole second.
• Sums durations when multiple videos are generated.
• Persists video duration in attachment metadata.
• Displays video duration in message cards, inline previews, file previews, and shared conversations.
• Prevents token, call, latency, or non-video requests from accidentally producing duration charges.
• Releases billing reservations when video generation fails or is canceled.
• Charges only results marked billable with a successful video-generation status.

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

• go test ./...
• pnpm check
• Targeted duration parsing and billing tests
• git diff --check

Configuration, migration, and compatibility notes

• No new environment variables or manual database migrations are required.
• Existing token-, call-, and tier-based pricing modes remain unchanged.
• Duration pricing can only be configured for models with the video_gen capability.
• Video duration defaults to 6 seconds when no usable duration is provided.
• Existing video attachments without duration metadata remain compatible.
• Failed, canceled, empty, or unavailable video results are not charged.

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 configuration, or personal data are included.
• Billing authorization and settlement remain scoped to the authenticated user.
• Failed video requests release their reserved billing budget.

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.