[Spanner Backup & Restore] Provide cell-level data integrity checking for backups

[EricBeach]

While we perform some data integrity checks now on the Spanner backups (e.g., validating table names and row counts), we do not validate any cell level data so it is theoretically possible that a bug in Pontem code or in the underlying Spanner/Dataflow connector code could lead to a cell at location 10, 2 being written instead to location 4, 56. In other words, there is currently no data integrity checking at the cell level. Using a hashing functionality or some other mechanism, the data should be checked at a cell level.