While investigating BF-43040 , colin.stolley@mongodb.com found SERVER-126023 which was able to improve performance on tsbs_bulk_load tests by ~15-30% by reordering some small boolean fields in OpDebug to avoid padding holes. (less holes = more compact layout = less memory used = less cache loads when loading fields into CPU cache). Clustering the bools or other smaller fields together utilizes memory that would otherwise be used by padding.

Since there might be some more perf wins from reordering fields here and OpDebug is a large struct, we should audit OpDebug for more quick wins here. For example, clustering smaller fields together, putting often-accessed fields together near the top (so you can avoid loading the entire struct into cache a lot of the time).