Remove redundant integration tests

[eurunuela]

Closes #258

Proposed Changes

• Remove redundant integration tests to gain in testing speed.

[eurunuela]

I don't know what's going on with Travis... some numpy related issues appeared and I honestly don't know why. Nothing has changed in the installation process... any idea @rmarkello ?

Anyway, the tests are much faster when removing the redundant ones (see here and they took 3:47 on my laptop). I checked codecov and we would have a -0.52% decrease in coverage (to 92.97%). The interfaces folder would still be covered at 98.83%, which means that removing the ACQ and TXT integration tests has no impact on txt.py and acq.py.

This means that the only ones below 90% would be bids.py (which will probably be above 90% when #254 is merged), due.py and phys2bids.py. I think it should be easy to increase coverage on them with silly unit tests without compromising testing speed.

So, I think merging this PR would improve our testing suite.

[vinferrer]

LGTM, i think this is ready for merge

[eurunuela]

We will have a bit of a decrease in overall coverage (down to 91%) but I expect #254 to increase that (bids.py from 73% to 98%).

[eurunuela]

Just to keep this written somewhere:

As per discussion over Zoom with @smoia and @vinferrer , I added the acq test back and merged the others into one single test (heuristic with multi-freq txt file and checks on the logger). The heuristic one takes 8 seconds on my laptop while the acq one takes around 40 seconds. I'll check if we have a short acq file on OSF and update if I find one.

Anyway, this is a very big decrease in testing time.

[vinferrer]

Let's merge