Spawning this from
HBASE-2375, I saw that it was much more efficient compaction-wise to check if we can split right after flushing. Much like the ideas that Jon spelled out in the description of that jira, the window is smaller because you don't have to compact and then split right away to only compact again when the daughters open.
Another thing it improves is while we're normally waiting for the compaction to happen, data that's still coming in will make us go way past the MAX_FILESIZE to a point where for the first region I was seeing a store size 3-4x bigger before it was able to split.
I targeted this for 0.94, but I'd like to get this into 0.92.1 or .2 too.