[Hardware] partially unrolled?

[John L. Bass]

For several reasons, the best choices are fully iterated, unroll one
stage, and fully unrolled. This is partially because the iterations
per stage are 13, and the SBox needs to be retimed in increments of
13 stages. Unrolled per stage, this is simply a FF. This will cause
some small problems with the L-terms, that will require a bit of
creativity, but manageable. Likewise the E terms need to be triggered
on the last round.

John