Playground: add video generation support and refactor token context setup

[LJR199887]

Motivation

• Add first-class video generation support in the Playground including UI controls, request building, and task fetching.
• Ensure token/context setup for playground routes is centralized and reused across chat and video flows.
• Extend request schema to include common video parameters so the relay can pass them upstream.

Description

• Introduce setupPlaygroundTokenContext to centralize writing user context and temporary token setup and update Playground to use it via Relay.
• Add PlaygroundVideoSubmit and PlaygroundVideoFetch handlers and register new routes POST /pg/video/generations and GET /pg/video/generations/:task_id in the router.
• Extend GeneralOpenAIRequest with seconds and quality fields to support video generation parameters.
• Frontend changes: add video controls to SettingsPanel (size, seconds, quality), add constants and defaults in playground.constants.js, and wire those fields into payload construction in helpers/api.js where video requests force non-streaming and set size, seconds, and quality.
• Update useApiRequest hook to detect video payloads, build a simplified video request payload from the last user message (extracting text and image), route requests to API_ENDPOINTS.VIDEO_GENERATIONS, handle non-stream responses for video tasks by summarizing task info into the message, and ensure streaming logic respects endpoints that must be non-streaming.

Testing

• Ran backend unit tests with go test ./... which completed successfully.
• Built the frontend with yarn build and ran type checks/linting, which succeeded.
• Performed integration smoke checks by exercising the new /pg/video/generations submit and fetch flows which returned expected task summaries in the Playground UI.

---

Codex Task