Return 500 on missing buckets in fallback layer (#8853)

### URL of deployed dev instance (used for testing):
- https://___.webknossos.xyz

### Steps to test:
- Add _ <- Fox.failure("test") to applyAgglomerate in AgglomerateService
- See 500 in zarr streaming response for a volume layer that has an
applied agglomerate.

The issue mentions:
"I would expect the MISSING_BUCKETS header to be set and zarr streaming
to return errors."
So should the MISSING_BUCKETS header also be added to the zarr streaming
response?

### TODOs:
- [ ] ...

### Issues:
- fixes #8667 

------
(Please delete unneeded items, merge only when none are left open)
- [x] Added changelog entry (create a `$PR_NUMBER.md` file in
`unreleased_changes` or use `./tools/create-changelog-entry.py`)

---------

Co-authored-by: Florian M <[email protected]>

Author: frcroth

unreleased_changes/8853.md

@@ -0,0 +1,2 @@
+### Changed
+- Zarr streaming a volume annotation will now respond with errors (500) if the underlying data encountered erros.

webknossos-tracingstore/app/com/scalableminds/webknossos/tracingstore/controllers/VolumeTracingZarrStreamingController.scala

@@ -352,7 +352,8 @@ class VolumeTracingZarrStreamingController @Inject()(
         (fallbackData, fallbackMissingBucketIndices) <- remoteDataStoreClient.getData(
           remoteFallbackLayer,
           List(request)) ~> SERVICE_UNAVAILABLE // return 503 if the request fails, assuming the datastore is still initializing
-        _ <- Fox.fromBool(fallbackMissingBucketIndices.isEmpty) ?~> "No data at coordinations in fallback layer" ~> NOT_FOUND // return 404 if the request worked but data is empty
+        // We only expect missing buckets if something went wrong with loading the fallback layer, e.g. applying the agglomerate. Otherwise, a fill value is used and returned to us.
+        _ <- Fox.fromBool(fallbackMissingBucketIndices.isEmpty) ?~> "Could not retrieve data from fallback layer" ~> INTERNAL_SERVER_ERROR
       } yield fallbackData
     } else Fox.successful(data)
 }