Match crawled listings to existing facilities (fuzzy) before creating new

When publishing a scraped listing we now look for a facility we already
have that is exactly or closely the same, and only create a new one when
there is no match — avoiding duplicates like «بیمارستان میلاد» vs «میلاد».

- ListingParser: extract a facility name (keyword + distinctive words) from
  the post and surface it in the parser notes.
- FacilityMatcher: Persian-aware normalization (ي/ك, ZWNJ, punctuation),
  type-word stripping for a "core" name, contains + Levenshtein similarity,
  and FindBest (same-city exact → any-city exact → same-city fuzzy → fuzzy).
- Review (manual publish): auto-select a matching facility or prefill the
  new-facility name; resolve-or-create uses fuzzy match; dropdown preselects.
- IngestionService (auto-publish): reuse FacilityMatcher against a run-wide
  facility list (grows as new ones are created) instead of exact-name only.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

src/JobsMedical.Web/Services/Scraping/IngestionService.cs

@@ -52,6 +52,7 @@ public class IngestionService
         var roles = await _db.Roles.ToListAsync(ct);
         var cities = await _db.Cities.ToListAsync(ct);
         var districts = await _db.Districts.ToListAsync(ct);
+        var facilities = await _db.Facilities.ToListAsync(ct);   // fuzzy-matched + grown as we create
         var roleNames = roles.Select(r => r.Name).ToList();
         var cityNames = cities.Select(c => c.Name).ToList();
         var districtNames = districts.Select(d => d.Name).ToList();
@@ -95,7 +96,7 @@ public class IngestionService
 
                 if (status == RawListingStatus.Normalized)
                 {
-                    try { Publish(parsed, ai, raw, roles, cities, districts); published++; }
+                    try { Publish(parsed, ai, raw, roles, cities, districts, facilities); published++; }
                     catch (Exception ex) { _log.LogWarning(ex, "Auto-publish failed; queueing instead"); raw.Status = RawListingStatus.New; queued++; }
                 }
                 else if (status == RawListingStatus.New) queued++;
@@ -157,7 +158,7 @@ public class IngestionService
     }
 
     private void Publish(ParsedListing parsed, AiAuditResult? ai, RawListing raw,
-        List<Role> roles, List<City> cities, List<District> districts)
+        List<Role> roles, List<City> cities, List<District> districts, List<Facility> facilities)
     {
         var d = ai?.Data;
         var roleName = d?.Role ?? parsed.RoleName;
@@ -170,9 +171,10 @@ public class IngestionService
         var district = districts.FirstOrDefault(x => x.Name == districtName && x.CityId == city.Id);
 
         var facilityName = !string.IsNullOrWhiteSpace(d?.FacilityName) ? d!.FacilityName!.Trim()
+            : !string.IsNullOrWhiteSpace(parsed.FacilityName) ? parsed.FacilityName!.Trim()
             : $"مرکز درمانی (از {raw.SourceChannel})";
-        var facility = _db.Facilities.Local.FirstOrDefault(f => f.Name == facilityName && f.CityId == city.Id)
-                       ?? _db.Facilities.FirstOrDefault(f => f.Name == facilityName && f.CityId == city.Id);
+        // Reuse an existing facility (exact or Persian-aware fuzzy match) before creating a new one.
+        var facility = FacilityMatcher.FindBest(facilities, facilityName, city.Id);
         if (facility is null)
         {
             facility = new Facility
@@ -181,6 +183,7 @@ public class IngestionService
                 Phone = parsed.Phone, IsVerified = false,
             };
             _db.Facilities.Add(facility);
+            facilities.Add(facility);   // so later listings in this run match it too
         }
 
         var kind = (d?.Kind ?? parsed.Kind.ToString()).ToLowerInvariant();