Test normality for small populations

[ksuchak1990]

Run StationSim_GCS for small population sizes and test normality of state vectors

[ksuchak1990]

Having run some instances of the EnKF for very small population sizes (2 agents and 5 agents) and ensemble sizes of 50, it doesn't look like state vectors are normally distributed. I have checked this by:

• Using histogram visualisations to look at the distribution of x and y coordinates of each agent in the population independent of each other, and
• pingouin.multivariate_normality() which implements the Henze-Zirkler test for multivariate normality.

Issues to raise with Jon:

• If the normality tests fail for such small population sizes, is there any chance that it will improve with larger population sizes? I don't think so.
• If the normality tests fail for the model in the absence of data assimilation, could this improve with data assimilation? Possibly?

[ksuchak1990]

Will review this in the morning, but this may be a dead-end/may not require that much more investigation