After cursor->modify() was enabled in mongoDB for DisAgg it caused a bunch of AFs:

• https://jira.mongodb.org/browse/AF-14327
• https://jira.mongodb.org/browse/AF-14705
• https://jira.mongodb.org/browse/AF-14380

After verifying that all these failures were caused by modify, we rolled back to always using insert for values updates in mongoDB (https://github.com/10gen/mongo/commit/b96cff0896ef8d946f52ccdff53d2d072690e743)

This brings a negative impact on performance, so we should enable modify() back when we fix the issues.

The minimal set of testing to enable modify is: 

• Enable test/format with modify (WT-16479)
• Run YCSB the same way it was executed on clusters from AF-14380 and check that with the new modify version it doesn't cause any failures at least for 2 days

Please note that after we disabled modify() we fixed a bunch of different bugs in layered cursors, so we should try to just do all the required testing first and see whether it'll cause any issues.

Another peculiar thing is that as I understand we disabled reserve() together with modify(), and now we know that reserve() was not working correctly at this point of a time (we have a PR that should fix this issue https://github.com/wiredtiger/wiredtiger/pull/13377 ) 

So it is worth checking whether reserve was the real reason for some of these AFs.