When we perform a selective PIT restore, we expect the caller to insert all oplog entries from the checkpoint timestamp to the point in time timestamp, regardless if the oplog entry references an unrestored collection. Normally, hitting an error during oplog application will crash the node. When `--restore` is used, we'll skip crashing on oplog application errors, since the oplog application error may be from a collection that we did not selectively restore.

It would be helpful to produce some log output here signaling that we skipped processing an oplog application error. This is something that could be backported safely to older versions, to help with debugging selective restores.