Add Server-Timing headers to all API endpoints for per-request latency breakdown

Instrument every API endpoint with Server-Timing headers so sub-operation
durations are visible in browser DevTools Network tab. Also adds Grafana
dashboard panels for per-endpoint latency comparison (p50/p95 timeseries
and p95 ranking bar gauge).

api/decision_routes.py

@@ -1,7 +1,8 @@
 import logging
+import time
 from typing import Annotated
 
-from fastapi import APIRouter, Depends, HTTPException, Query
+from fastapi import APIRouter, Depends, HTTPException, Query, Response
 from pydantic import BaseModel, Field
 
 from api.auth import User, get_current_user
@@ -55,11 +56,17 @@ async def set_decision(
     user: Annotated[User, Depends(get_current_user)],
     listing_id: int,
     body: SetDecisionRequest,
+    response: Response,
 ) -> DecisionResponse:
     """Set or update a like/dislike decision for a listing."""
+    timings: list[str] = []
+    t0_total = time.monotonic()
+    t0 = time.monotonic()
     user_id = _get_user_id(user)
+    timings.append(f"user_lookup;dur={(time.monotonic() - t0) * 1000:.1f}")
     repo = DecisionRepository(engine)
     try:
+        t0 = time.monotonic()
         result = decision_service.set_decision(
             repo,
             user_id=user_id,
@@ -67,19 +74,31 @@ async def set_decision(
             listing_type=body.listing_type,
             decision=body.decision,
         )
+        timings.append(f"upsert;dur={(time.monotonic() - t0) * 1000:.1f}")
     except ValueError as e:
         raise HTTPException(status_code=400, detail=str(e))
+    timings.append(f"total;dur={(time.monotonic() - t0_total) * 1000:.1f}")
+    response.headers["Server-Timing"] = ", ".join(timings)
     return _to_response(result)
 
 
 @decision_router.get("", response_model=list[DecisionResponse])
 async def get_decisions(
     user: Annotated[User, Depends(get_current_user)],
+    response: Response,
 ) -> list[DecisionResponse]:
     """Get all decisions for the current user."""
+    timings: list[str] = []
+    t0_total = time.monotonic()
+    t0 = time.monotonic()
     user_id = _get_user_id(user)
+    timings.append(f"user_lookup;dur={(time.monotonic() - t0) * 1000:.1f}")
     repo = DecisionRepository(engine)
+    t0 = time.monotonic()
     decisions = decision_service.get_user_decisions(repo, user_id)
+    timings.append(f"fetch;dur={(time.monotonic() - t0) * 1000:.1f}")
+    timings.append(f"total;dur={(time.monotonic() - t0_total) * 1000:.1f}")
+    response.headers["Server-Timing"] = ", ".join(timings)
     return [_to_response(d) for d in decisions]
 
 
@@ -87,17 +106,26 @@ async def get_decisions(
 async def delete_decision(
     user: Annotated[User, Depends(get_current_user)],
     listing_id: int,
+    response: Response,
     listing_type: str = Query(..., description="RENT or BUY"),
 ) -> dict[str, bool]:
     """Remove a decision (un-like/un-dislike)."""
+    timings: list[str] = []
+    t0_total = time.monotonic()
+    t0 = time.monotonic()
     user_id = _get_user_id(user)
+    timings.append(f"user_lookup;dur={(time.monotonic() - t0) * 1000:.1f}")
     repo = DecisionRepository(engine)
     try:
+        t0 = time.monotonic()
         deleted = decision_service.remove_decision(
             repo, user_id, listing_id, listing_type
         )
+        timings.append(f"delete;dur={(time.monotonic() - t0) * 1000:.1f}")
     except ValueError as e:
         raise HTTPException(status_code=400, detail=str(e))
     if not deleted:
         raise HTTPException(status_code=404, detail="Decision not found")
+    timings.append(f"total;dur={(time.monotonic() - t0_total) * 1000:.1f}")
+    response.headers["Server-Timing"] = ", ".join(timings)
     return {"success": True}