Create a brief from a user-pasted text brief (Entry #2)

Inserts the brief in `queued` status and enqueues the parser pipeline asynchronously, returning the briefId in <100ms. The pipeline parses the raw text via Claude Sonnet (outline_extraction → enrichment), populates the brief document, and lands the brief in `brief_editing`. Skips the 14-step research pipeline and does NOT auto-start content generation — the caller (UI or agent) must walk the brief editor and trigger content generation through `POST /domains/:domainId/content-briefs/:briefId/content/generate`. The full CONTENT_GENERATION credit cost is checked there, but this endpoint also runs a pre-flight CONTENT_COST_MIN floor check so callers without the minimum content credits fail fast (mirrors the AI flow's BRIEF_GENERATION pre-flight) — that is what produces a 402 here. Subscribe to `subscribe_content_brief` socket events with the returned briefId to follow parser progress.