SERVER-126514 adds some oplog convergence checks to pali_chaos.js to make sure oplog truncation works after lots of failover events. The maxSize assertion that was added was pretty relaxed (size < 8 * maxSize) to get the tests to pass on evergreen; we should spend some time investigating why the oplog growth is so much more the maxSize.

In particular, we should confirm that this is happening because of the extreme test setup (low maxSize and frequent restarts) and not something that could show up realistically in production.

Some possible avenues to explore to narrow the 8x number are: setting the maxSize to be larger from 5 MB (may have to adjust the oplog generated in the tests), setting waits to let the oplog to converge before checking maxSize, checking if there are pins or anything else that are holding the truncation point back.

This also seems related to SERVER-122180 (worst case oplog growth), so we might want to confirm worst case oplog growth on ASC