Upon restart, if a node has a prepared internal transaction for retryable findAndModify executed with storeFindAndModifyImagesInSideCollection=true to recover, the reconstruct prepared transactions step would get stuck indefinitely as applying the prepare oplog entry appears to keep hitting WriteConflict errors. This error can be reproduced using these two test cases in retryable_findAndModify_commit_and_abort_prepared_txns_after_failover.js by making the jstest trigger a restart instead of a failover.