Support masking of arrays with nans for relaxation #158
Description
Currently for the MSD I am only using values between 2 and 100 for the calculation of the diffusion constant. This is fine for translational diffusion, however rotational diffusion I would like to have a smaller maximum value.
The simplest fix to this is probably the handling of nan values within the calculation, dropping them appropriately. This would mean the masking can occur from the caller, not requiring any logic or change within this function.