Summary:
Different names were used for number of rows read from the table
during backfill such as `num_rows_processed`, `number_rows_processed`,
`rows_processed_entries` and so on. These are renamed to
`num_rows_read_from_table_for_backfill` for clarity.

Number of rows read during each batch were being accumulated in an
atomic field in `BackfillTable::UpdateRowsProcessedForIndexTable` and
then being set in the `BackfillJobPB`. This is unnecessary since this block is
protected by a write lock on the indexed table. Instead, we can directly
increment the value in the backfill job.

### Upgrade/Downgrade safety:
Only names of fields in protobuf are changed, but the field numbers
remain the same. This change should be safe for upgrade/downgrade.
Jira: DB-18811

Test Plan: Jenkins

Reviewers: smishra, jason, amitanand

Reviewed By: jason

Subscribers: ybase, yql

Differential Revision: https://phorge.dev.yugabyte.com/D47699