perf(db): Count hits on a replica instead of the primary postgres (#32178)

This can be an expensive query, and it doesn't need to be 100% accurate, so we can safely use a
replica here. Replag is typically low anyway, so this shouldn't have any noticeable customer impact.

Author: wedamija

src/sentry/api/paginator.py

@@ -190,7 +190,7 @@ def count_hits(self, max_hits):
             h_sql, h_params = hits_query.sql_with_params()
         except EmptyResultSet:
             return 0
-        cursor = connections[self.queryset.db].cursor()
+        cursor = connections[self.queryset.using_replica().db].cursor()
         cursor.execute(f"SELECT COUNT(*) FROM ({h_sql}) as t", h_params)
         return cursor.fetchone()[0]