feat: Add auto-trigger for LLM analysis

Introduces a new module `auto_trigger` responsible for opportunistically
initiating LLM analysis jobs. This mechanism scans case directories for
new or updated recordings and transcripts, automatically creating and
queuing analysis jobs when appropriate.

Key components:
- `evaluate_case`: Pure function to determine if a case is ready for
  analysis based on recording status, transcriptions, document
  modification times, and existing job/failure markers.
- `try_enqueue`: Orchestrates the analysis job creation and queuing
  process if `evaluate_case` returns `AutoDecision::Enqueue`.
- `try_enqueue_all_for_user`: Iterates through all user cases to trigger
  `try_enqueue` for eligible ones.
- `write_failure_marker`/`remove_failure_marker`: Handles persistent
  recording of analysis failures and their cleanup.

This feature aims to reduce manual intervention by automatically
analyzing new or changed case data.

server/templates/case_page.html

@@ -157,6 +157,10 @@ try {
     timer = setTimeout(() => location.reload(), 300);
   };
   const es = new EventSource('/web/events');
+  // Release the socket-pool slot on navigation. Without this, Chrome
+  // keeps the SSE connection "draining" after unload; six rapid nav
+  // cycles exhaust the 6-per-origin pool and stall further requests.
+  window.addEventListener('pagehide', () => es.close());
   es.addEventListener('case', (msg) => {
     let evt;
     try { evt = JSON.parse(msg.data); } catch (_) { return; }