Comments above _bt_findinsertloc() that talk about LP_DEAD items are now out of place. We already discuss index tuple deletion at an earlier point in the same comment block. Oversight in commit d168b666.