Run manager parameters

[sdhiscocks]

In discussion with @gbellantuono, wondering what types of parameters that run manager should support.

So basic types, like int (e.g. number of particles) and float (e.g. noise coefficients). Boolean as well.

I'm not sure where str would be useful, as more than likely if it is a parameter to change for example the choice of an internal method, this should be a Enum.

datetime.timedelta would be useful, as a parameter that is used in many places. This could simply be represented as seconds, such acts like float with min/max, etc.

For things like, tuple and list and NumPy arrays, in particular StateVector and CovarianceMatrix types, you may only want to vary a subset of the values, or all of them. I think CovarianceMatrix has a special case where you may only want to vary the diagonal elements.

I can also see use case where you may want to change an entire component, e.g. swapping out two different transition models. However, this may also require changing out other variables (e.g. change from CV to CA model, you'll need to modify prior to include acceleration). Considering these as completely different configurations may be easier initially.

When varying parameters, I could see it useful to have a way to choose how that value varies. Simple case is a min and max value, with a number of steps, or step size. But what also would be useful is random sampling, or quasi-random sampling. I can also see use case where you may have min and max value, but may want to step in log/exp scale.