... break-out of the padded cell of prove and run subtests individually ... isn't that what I did when I responded to tye's suggestion ("perl -Mblib t/02parse.t")?

Yes, you were on the right track there, and I would encourage you to continue to pursue that general strategy.