Prepare workflow APIs for compose advisory integration

[rockythorn]

Summary

• Replace the asyncio.wait_for(handle.result(), timeout=1.0) status heuristic with Temporal's describe() API, which returns a definitive status enum (RUNNING, COMPLETED, FAILED, etc.) plus start/close times
• Use deterministic workflow IDs (rh-matcher-compose-v{major}) for single-version triggers so Temporal deduplicates concurrent runs via id_reuse_policy=ALLOW_DUPLICATE
• Return 409 Conflict when a trigger is attempted for a version that already has a running workflow
• Add live workflow status panel to the admin UI with 15-second polling after triggering a workflow

These changes prepare the Apollo server for compose pipeline integration, where trigger-advisory-clone.sh (toolkit side) will call the trigger and status endpoints to ensure fresh advisory data before apollo_tree generates updateinfo.xml.

Test plan

• Trigger endpoint returns deterministic ID rh-matcher-compose-v9 for {"major_versions": [9]}
• Status endpoint returns running with start_time, run_id via describe() API
• Duplicate trigger while workflow is running returns 409
• Multi-version and no-version triggers still use UUID-based IDs (backward compat)
• Admin UI status panel polls and updates after triggering a workflow
• Admin UI shows "already running" error on duplicate trigger