fix(admin): LLM provider/model switching now reads DB overrides correctly

- get_llm_provider() now accepts provider + model args so DB overrides
  propagate through to all provider constructors (was always reading
  env vars, ignoring the admin UI selection)
- /test-llm replaced with lightweight ping (max_tokens=20) instead of
  running a full bill analysis; shows model name + reply, no truncation
- /api/settings/llm-models endpoint fetches available models live from
  each provider's API (OpenAI, Anthropic REST, Gemini, Ollama)
- Admin UI model picker dynamically populated from provider API;
  falls back to manual text input on error; Custom model name option kept
- Default Gemini model updated: gemini-1.5-pro → gemini-2.0-flash

Co-Authored-By: Jack Levy

backend/app/workers/llm_processor.py

@@ -60,7 +60,13 @@ def process_document_with_llm(self, document_id: int):
             .first()
         )
 
-        provider = get_llm_provider()
+        from app.models.setting import AppSetting
+        prov_row = db.get(AppSetting, "llm_provider")
+        model_row = db.get(AppSetting, "llm_model")
+        provider = get_llm_provider(
+            prov_row.value if prov_row else None,
+            model_row.value if model_row else None,
+        )
 
         if previous_full_brief and previous_full_brief.document_id:
             # New version of a bill we've already analyzed — generate amendment brief
@@ -97,6 +103,9 @@ def process_document_with_llm(self, document_id: int):
 
         logger.info(f"{brief_type.capitalize()} brief {db_brief.id} created for bill {doc.bill_id} using {brief.llm_provider}/{brief.llm_model}")
 
+        # Emit notification events for users who follow this bill
+        _emit_notification_events(db, bill, doc.bill_id, brief_type, brief.summary)
+
         # Trigger news fetch now that we have topic tags
         from app.workers.news_fetcher import fetch_news_for_bill
         fetch_news_for_bill.delay(doc.bill_id)
@@ -111,6 +120,35 @@ def process_document_with_llm(self, document_id: int):
         db.close()
 
 
+def _emit_notification_events(db, bill, bill_id: str, brief_type: str, summary: str | None) -> None:
+    """Create a NotificationEvent row for every user following this bill."""
+    from app.models.follow import Follow
+    from app.models.notification import NotificationEvent
+    from app.config import settings
+
+    followers = db.query(Follow).filter_by(follow_type="bill", follow_value=bill_id).all()
+    if not followers:
+        return
+
+    base_url = (settings.PUBLIC_URL or settings.LOCAL_URL).rstrip("/")
+    payload = {
+        "bill_title": bill.short_title or bill.title or "",
+        "bill_label": f"{bill.bill_type.upper()} {bill.bill_number}",
+        "brief_summary": (summary or "")[:300],
+        "bill_url": f"{base_url}/bills/{bill_id}",
+    }
+    event_type = "new_amendment" if brief_type == "amendment" else "new_document"
+
+    for follow in followers:
+        db.add(NotificationEvent(
+            user_id=follow.user_id,
+            bill_id=bill_id,
+            event_type=event_type,
+            payload=payload,
+        ))
+    db.commit()
+
+
 @celery_app.task(bind=True, name="app.workers.llm_processor.backfill_brief_citations")
 def backfill_brief_citations(self):
     """