Update Overlap Drop Parameter (small algorithmic update) about fooof

[voytek]

Without sitting with the data more, it's hard for me to say. I just want to make sure you're not over-tuning to MEG. Have you tried performance on EEG and LFP to see what's better for those?

[TomDonoghue]

I scanned through the EEG & LFP data I have available as well.

It's a relatively minor change, in that it has no effect whenever multiple gaussians are not near each other, so for many fits, this param is not invoked.

Since we don't have an objective manner to set parameters like this, it's all a bit arbitrary, but my best-guess update is that it's better to push away from overlapping gaussians a bit more than we do currently (currently my impression is it ends up with unhelpful overlaps).

So, I've updated to 1.5.

We can revisit this (and other parameters) a bit later as we get more of a sense of how the algorithm is working across different datasets.