Spark: Support Bulk deletion in expire-snapshots if fileIO allows

[dramaticlly]

In #4052, S3 fileIO now implement a new interfaces to support S3 batch deletion, this PR introduce it for Spark expire-snapshots procedure to conditionally support delete files in batch if underlying fileIO supports it (if implements SupportsBulkOperations interface and currently only S3FileIO support such)

It default to use S3 batch deletion if fileIO is supported in catalog, allow for customization with bulkDeleteWith method

• cannot reuse the existing bulkDelete consumer function because it only take single file name at a time instead of a iterable
• did not add interface override, want to keep this change as small as possible, once approved then we can retroactively apply to interface and previous spark version (2.4/3.0/3.1/3.2)
• Considering the existing test fixture all use HadoopTables and it's very hard to test the integration of S3FileIO and Spark action together in unit tests, so I am looking for a way to do some integration tests and will share the results later

Similar to #5373 but for expire-snapshots procedure, relate to #4012

CC @rdblue , @danielcweeks, @amogh-jahagirdar, @szehon-ho

[dramaticlly]

I shall probably move this table.io() to ops.io() for consistency, but given check take 1+hr to complete, I will handle it together with other review feedback

[szehon-ho]

Probably we need to synchronize with @amogh-jahagirdar on whether we need a bulkDeleteFunc, and whether retry happens at S3FileIO layer or not.

[dramaticlly]

closed in favor of #6682, which cover more comprehensive spark actions and also implement bulk deletion for HadoopFileIO