Updated clean_strain() to return spike model when return_mcmcs is tru…

[HarrisonS-Phys]

…e, and removed extraneous lines of code

[farr]

I'm curious about the return values here---why don't you return the time-domain line model instead of the real and imag freq components? (Just thinking about Fourier transform conventions---everything input to the cleaner is real, and if everything output from the cleaner is also real, then we don't have to fuss about definitions of FTs, because it's all internal; as well, it's easier to understand the time-domain line model, no?)

Also: why are you just returning the real part of the data frequency series? Sort of the same question?

[farr]

Also, also: isn't there an extra s in the return pred_sampless? Does this code path even work? :cringe:

[farr]

Finally, finally (I hope): note that you can extract the frequency components of the line model already from the predicted samples---so maybe it would be better just to store the td line model in the prediction instead of returning separately?