[Beowulf] MD check/scrub

[Mark Hahn]

> If they find an inconsistent stripe they don't try to identify the
> corrupt block. Instead, they dumbly *recompute P and Q*, which of
> course makes the stripe consistent, but *leaves the corrupt data in
> place*.

I'm skeptical that MD would be so willfully stupid.  did you conclude
this from reading the code, empirical testing, or from the author?
to me, the code looks to be doing something sensible (it definitely
acts differently when the failure(s) are on P/Q/data.)