fix(ui): use Govmap resolved address text for street inference, not frequency heuristic

Previous approach: pick most common streetNameHeb among street-source deals.
Problem: a busier nearby street wins even when the searched parcel is on a
quieter street (e.g. רב צעיר beats הילדסהיימר despite being a different street).

Fix: backend now returns resolved_address from Govmap autocomplete (e.g.
"הילדסהיימר 14, ירושלים"). Frontend parses the leading non-numeric tokens
as the street name and uses that as ground truth for proximity labeling.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

nadlan_mcp/web_app.py

@@ -165,13 +165,15 @@ def create_app() -> FastAPI:
             logger.exception("find_recent_deals_for_address failed")
             raise HTTPException(status_code=502, detail=str(e))
 
-        # Search-center coords for the map (best-effort).
+        # Search-center coords + resolved address text for the UI (best-effort).
         search_coords = None
+        resolved_address = None
         try:
             ar = govmap_client.autocomplete_address(req.address)
             if ar.results and ar.results[0].coordinates:
                 c = ar.results[0].coordinates
                 search_coords = {"longitude": c.longitude, "latitude": c.latitude}
+                resolved_address = ar.results[0].text
         except Exception:
             pass
 
@@ -198,6 +200,7 @@ def create_app() -> FastAPI:
         return {
             "search": {
                 "address": req.address,
+                "resolved_address": resolved_address,
                 "coordinates": search_coords,
                 "years_back": req.years_back,
                 "radius_meters": req.radius_meters,