Find reasonable default for ExponentialNoise tau parameter

[jschepers]

I think the proper way would be to estimate intrinsic times via #161 of real EEG data and then recommend a parameter here

Originally posted by @behinger in #129 (comment)

[behinger]

based on yasir calal's phd thesis / paper he published, I found intrinsic timescale measures of 0.2-0.4 for ACW0 for EEG data.

one can estimate a relationship between our tau and their acw-0 which I did here: https://pluto.land/n/cmq1kt1i

Reasonable tau-values seem to be 0.03-0.0.04 * fs, so by default fs of 100, $\tau$ of 3-4.

But, the tau#s are also influenced on the actual ERPs one simulates, I think because they simply add a different type of autocorrelation themselves.

This is just one hint, I didnt really look into autocorrelation functions / PSDs and tried to compare