backupccl: allow SSTBatcher to split ranges during RESTORE

[msbutler]

Currently, during RESTORE, the SSTBatcher is not configured to split a range it's ingesting. This isn't a big deal because the ranges RESTORE creates and writes data to tend to be under-filled. Specifically, a single SSTBatcher in RESTORE ingests a single restoreSpanEntry's data into a single range, but the amount of data associated with a restoreSpanEntry tends to be much smaller than the default range size. However, once #83139 implements restoreSpanEntry target sizes, an SSTBatcher may fill multiple ranges worth of data. To prevent post restore splits, the SSTBatcher should split the range its ingesting once it has ingested a range's worth of data.

Since the SSTBatcher operates on a specific node, these SSTBatcher splits would also prevent any post-Restore splits on a cluster that has zone configs that specify different target range sizes for different zones.

Jira issue: CRDB-16887

Epic CRDB-10127

[blathers-crl]

cc @cockroachdb/bulk-io

[msbutler]

Fixed by #86496