Lotka-Volterra ODE semantics for causal loop diagrams with indeterminates

[epatters]

Sequel to #206 and #111. The idea is that the numerical parameters associated with positive/negative links should have fixed signs (as for ordinary causal loop diagrams) but the parameters associated with indeterminate links can have arbitrary sign. So the semantics for indeterminates is less constrained, as one would expect.

It should be possible to implement this feature with minimal code duplication.