Have you also considered making it difficult to create a large sized cell without a nucleus? For example a typical human skin cell has 30 micrometers of diameter and a standard E.Coli 0.6 micrometers (mitochondria have a diameter around 0.8 micrometers; E.Coli and mitochondria have similar lengths, overall they can be considered to be about the same size). So the defined size of an Hexe defines the multiplier, but following the 90% multiplier, for example:
too many hexes = 11 * multiplier / (1 - multiplier)
so the number of hexes beyond the minimum to break even with a nucleus would be simply:
difference = multiplier / (1 - multiplier)
considering 90%
difference = 9
so there could be an osmoregulation penalty (or some other, e.g. max health, absorption rate) when a cell has over 100 hexes without a nucleus.
On the hand, if considering that, currently, the mitochondria is around 1.5 μm2 (equivalent to 2 hexes) and a human skin cell is aproximately 40 μm2, it would be 53 hexes. If considering this as the point where having a nucleus breaks even (which would be a very rough approximation) the multiplier would be 84%, thus the maximum number of hexes without a nucleus without suffering penalties would be 58.