[ENH] - Tech Audit #2: Accuracy checks for Hilbert Transform, Filters, and Spectra

[elybrand]

This PR adds accuracy checks (#194) for

1. All hilbert.py related functions using a simple sinusoid.
2. FIR filtering of a low-frequency sinusoid. Specifically, it checks that it is attenuated under a high pass filter, and is minimally attenuated under a lowpass filter.
3. compute_spectrum with the medfilt and welch options. It uses a low frequency sinusoid to compare against where the power spectrum should be a dirac spike at frequency 1 and zero elsewhere.

[ryanhammonds]

These tests updates look good to me!

[TomDonoghue]

Okay, slightly unconventional review approach here. I thought I had just some small updates, so I started doing directly. Then I realized I had a bit more to suggest - which I thought I would add as comments, rather than digging in too deep with direct edits. So, in the end I did pushed an update commit, and I'm also leaving some review comments.

Please take a look at my direct commit, to sanity check, and get a sense of what I was going for. Basically, that commit is trying to consolidate on using general test settings and refactoring some code a bit.

[TomDonoghue]

@elybrand - thanks for the updates here! These look much better / more specific tests than we had before. I've made some comments / suggestions. A bunch of these things are more code mechanics, so if you prefer you can tag in @ryanhammonds to work on the code and organizational elements here, and we can coordinate this one as a team effort.

I also wasn't totally sure if this PR was done, or a work-in-progress (if you plan on adding more test updates). Note that if things are in progress, you can use the tag [WIP] in the title, so that everyone knows there's more coming.

[TomDonoghue]

This looks great @elybrand! I've made some small comments for minor organizational things. Once you get a chance to check and update those, then I think this should be all good!

[TomDonoghue]

Looks good to me, merging in!