Organize blobs on disk by epoch (#14023)

* organize blob directories by period and epoch

* changelog

* remove Indices and replace with Summary

* old PR feedback

* log to advise about the speed of blob migration

* rename level->layer (hoping term is more clear)

* assert path in tests for increased legibility

* lint

* lint

* remove test covering a newly impossible error

* improve feedback from flag validation failure

* Try to clean dangling dirs epoch->flat migration

* lint

* Preston feedback

* try all layouts and short-circuit if base not found

---------

Co-authored-by: Kasey Kirkham <kasey@users.noreply.github.com>

beacon-chain/das/availability.go

@@ -94,14 +94,7 @@ func (s *LazilyPersistentStore) IsDataAvailable(ctx context.Context, current pri
 	entry := s.cache.ensure(key)
 	defer s.cache.delete(key)
 	root := b.Root()
-	sumz, err := s.store.WaitForSummarizer(ctx)
-	if err != nil {
-		log.WithField("root", fmt.Sprintf("%#x", b.Root())).
-			WithError(err).
-			Debug("Failed to receive BlobStorageSummarizer within IsDataAvailable")
-	} else {
-		entry.setDiskSummary(sumz.Summary(root))
-	}
+	entry.setDiskSummary(s.store.Summary(root))
 
 	// Verify we have all the expected sidecars, and fail fast if any are missing or inconsistent.
 	// We don't try to salvage problematic batches because this indicates a misbehaving peer and we'd rather