We would like to either log a message or increase a statistic if we hit the failure case that modifying the update chain fails after we have successfully inserted the updates onto the update chain.

This is a follow on ticket of WT-6763, in which we discovered an edge case that after we have successfully inserted updates onto the update chain, we can still fail in __wt_txn_modify. In the failure case, we freed the updates already inserted onto the update chain in error handling, which causes a read after free issue.

We should monitor how often such case happen.