In WT-8691, we removed the key_consistent write timestamp usage keyword.

A side-effect of that change was to enable more key-consistency checking for the ordered keyword than before, which broke the MDB build.

WT-8748 reverted the change to add additional checking, this ticket is a reminder to add the additional checking back in to MDB server builds if/when that's possible.

The change is in txn/txn.c, where we now only run some of the key-ordering checks if the WT_STANDALONE_BUILD flag is defined, which limits the checks to WiredTiger's internal testing, and not MDB server runs. See the WT-8478 commits for the details.