Code review comment for lp:~vorlon/lp-ftbfs-report/report-only-regressions

ok, the "only reported for armhf" is because (not obviously to me) the results of previous runs are cached in a way that means if you change the list of architectures you're running it for, it doesn't properly update the results.

So the only remaining thing I'm looking at is to check the runtime impact of the added checks.