feat(search): configure Typesense for Vietnamese diacritic search

Add normalized (ASCII-only) fields to Typesense schema and indexer so
users can search without diacritics (e.g. "can ho" finds "căn hộ").
Create synonym collection for HCMC district abbreviations and common
property-type aliases. Enable num_typos:2 for fuzzy matching.

- Add 7 normalized fields (title, description, address, ward, district,
  city, projectName) using Address.normalize() at index time
- Search queries both original Vietnamese and normalized field sets
- Upsert 28 Vietnamese synonym rules on collection init
- Normalize user query to ASCII alongside original for dual matching
- Update tests for new fields and synonym upsert behavior

Co-Authored-By: Paperclip <noreply@paperclip.ing>

apps/api/src/modules/search/infrastructure/services/search-result-mapper.ts

@@ -1,4 +1,5 @@
 import { type ListingDocument } from '../../domain/repositories/search.repository';
+import { Address } from '@modules/listings/domain/value-objects/address.vo';
 
 export interface RawListingRow {
   listingId: string;
@@ -64,5 +65,14 @@ export function mapRowToListingDocument(row: RawListingRow): ListingDocument {
     legalStatus: row.legalStatus ?? null,
     amenities: Array.isArray(row.amenities) ? (row.amenities as string[]) : [],
     isFeatured: row.featuredUntil && new Date(row.featuredUntil) > new Date() ? 1 : 0,
+
+    // Vietnamese diacritic-normalized fields
+    titleNormalized: Address.normalize(row.title),
+    descriptionNormalized: Address.normalize(row.description),
+    addressNormalized: Address.normalize(row.address),
+    wardNormalized: Address.normalize(row.ward),
+    districtNormalized: Address.normalize(row.district),
+    cityNormalized: Address.normalize(row.city),
+    projectNameNormalized: row.projectName ? Address.normalize(row.projectName) : null,
   };
 }