Allow expliclitly defining arbitrary patch statistics via lookup table

[bruno-f-cruz]

P_dependent = [*arg]

Independent variables:

• choices -> easy, just use the integer value as an index in the array
• reward -> Can easily deadlock if P_reward[0] = 0
• time, assume index = 1second and linearly interpolate
• distance, assume distance = 1cm and linearly interpolate

Boundary condition? if i > len(P_dependent) => P_dependent[-1]

[bruno-f-cruz]

Should probably re-use some sort of implementation for this too:

https://github.com/AllenNeuralDynamics/Aind.Behavior.Services/pull/67/files

[bruno-f-cruz]

A more general solution is to allow users to define a List<List> that can be used as a LUT where X[0] and X[1] encode the dependent and independent variables. Values in between fallow the same logic as it is currently implemented using available dependent variables, but the code will interpolate between the values in X.

Discrete cases will be trivially encoded by passing an exhaustive list of all values that the patch should transition through.