over-engineering audit: remove dead code, speculative config, duplicate queues

- delete embedded community resolver snapshot (6,760 lines) + EmbeddedServers +
  DNS_SERVERS=fresh; curated resolvers are the default. Simplify Warm (no
  sampling).
- one generic internal/sample.Queue[T] replaces three near-identical
  random-sample queues (probe, pipeline, discover).
- delete unreachable index.listOpenSearch (OpenSearch is never listable).
- delete classifyCloudFront (its match is always discarded, !ResourceSpecific).
- delete probe.Probe wrapper + Describe; tests use ProbeDetailed.
- delete store.Pool(), model.SourceDNSAlias, discover.QueueLen.
- drop dead config/no-op knobs: pipeline DirectWorkers, discover MaxPerDomain,
  discover S3/GCS/DO flags, BUCKET_SCAN_DO.
~-7,000 lines (mostly the resolver snapshot), 0 deps removed, tests green.

internal/store/store.go

@@ -47,8 +47,6 @@ func (s *Store) Ping(ctx context.Context) error { return s.pool.Ping(ctx) }
 
 func (s *Store) Close() { s.pool.Close() }
 
-func (s *Store) Pool() *pgxpool.Pool { return s.pool }
-
 // PoolStats reports pgx pool utilization for logging/metrics.
 func (s *Store) PoolStats() (total, idle, acquired int32) {
 	st := s.pool.Stat()