HDDS-14467. Expose pending delete bytes/namespace in OzoneBucket.

[sadanand48]

What changes were proposed in this pull request?

After HDDS-13756, OzoneBucket.usedBytes is populated from OmBucketInfo.getTotalBucketSpace() (used + snapshotUsed bytes), which makes bucket usage ambiguous after key deletion. This change adds explicit pendingDeleteBytes / pendingDeleteNamespace fields to OzoneBucket to clarify retained bytes (snapshots or pending delete).

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-14467

How was this patch tested?

Unit tests

[smengcl]

Thanks @sadanand48 for the patch.

Let's document the behavior of PendingDeleteBytes / PendingDeleteNamespace (e.g. not accounting for keys that are not recursed yet).

[sadanand48]

created https://issues.apache.org/jira/browse/HDDS-14642 for the same.

[sadanand48]

@jojochuang @smengcl FYI.
For recursive deletes in FSO buckets, pendingDeleteBytes / snapshotUsedBytes are not updated when the directory is deleted. They are updated only when DirectoryDeletingService moves subfiles into the deleted table via PurgeDirectoriesRequest (handled by OMDirectoriesPurgeRequestWithFSO), which calls decrUsedBytes(..., true) and increments snapshotUsedBytes.

I think this should be okay as even today the updates to usedBytes and usedNamespace happens in the same way.

Exposing snapshotUsedBytes might still be useful when snapshot lifespan is significant, as it helps users understand quota usage and why space is not reclaimed after deletes.