We've relied on this property in the past, to prove to an installation that the database corruption they were experiencing wasn't an LMDB bug - it turned out to be a faulty DRAM DIMM in their server. Without LMDB's deterministic write behavior, convincing someone of these cases would be a lot more difficult.
But proving the presence of a bug and narrowing down its cause are really two different tasks, and thus requires two different types of logging.