Mystery solved; zero FAILs again with mainline on freebsd5

[Loren James Rittle]

>For consistency, I think would be nice changing at the same time the 
>plain char counterparts of those tests...

If we find a host which can't handle that stack size, I'd agree.  The
tests do not have to be the same.  It was an easy problem to find once
I looked for it.  If it were a hard problem to find, then I'd consider
switching them all in a second. - Loren

PS, I don't have a strong negative to aligning them now that you
observed the parallel (the above banter was in regards to how hard I'd
look to find places to change).