Alright, here's my first stab at reactor design:
http://www.talonfiremage.pwp.b…5fpra56x3p93qzfw568o8nb40 (Not the link I had planned, that was a T-shape, but has the same stats. Uglier, though D:
It's somewhere between reactors 2 and 4 on the one-cycle Mark II reactor list, having higher EU/t than reactor 2 and better efficiency than reactor 4 (however, it has higher cooldown than both). I'm only bothering to post it because it uses five cells and those two use 4 and 6 respectively, so anyone wishing to use five cells and a setup similar to that can use this design. I'll keep tooling around with five cell (since 4cell seems to have hit a dead end as described below and 3cell has been solved for perfect generation) and see what I can come up with.
I also suspect that 4-cell reactors (all cells connected to each other, not in a box shape) can get to a minimum of 5 heat generated per tick for mathematical reasons. I keep trying to get lower and it just does not work, and it seems like it's a combination of the reactor size and the way crosses tessellate in two dimensions. I'm not going to go and prove it, though.