Minor fix to stats code; Watch for a swapmod on a swapped out
experiment, and not count its pnodes or duration as if it was swapped in! Thanks Kirk. The actual fix is simple. The script to fix up the existing stats records was a bear and took me all morning to get right! Even so, its probably no perfect, but close enough. That script in the sql directory and mentioned in doc/UPDATING.
Showing with 142 additions and 3 deletions