empirically choose better count of K

[dave-doty]

Currently we set K to be n, the number of molecules in the original CRN at the start of a batch, whenever n changes sufficiently. This choice can affect the number of passive reactions, reducing efficiency. Here's a plot from the paper of the fraction of passive reactions in the Lotka-Volterra oscillator:

It make be possible to reduce the fraction of passive reactions through a more careful choice of K.

[JoshPetrack]

This can be analyzed using theory - there is some function of K count that probably has a unique global minimum, describing the expected fraction of passive reactions. Could be theoretical reasons to set it to either side of that though so ultimately will need to be picked empirically.