Fixes for failing Travis tests

[dnowacki-usgs]

@rsignell-usgs I committed a few PRs yesterday, in particular python3 updates (#314) and unpinning dependency versions (#316). Before, conda couldn't even solve the dependencies and would fail at the get go. Now, everything passes except for one test... I'm not sure, but it might have something to do with masked arrays. The arrays at a quick glance look mostly equivalent--at least the first rows do, but one is masked and one isn't. Any ideas?

https://travis-ci.org/USGS-CMG/usgs-cmg-portal/jobs/659163616#L3264

[rsignell-usgs]

Seems like something changed and we are now getting masked arrays, so need to change the assert, right?

[dnowacki-usgs]

I looked closer and for some reason u_1205 in the EUROSTRATAFORM test file is coming in as a fully masked array (the mask is all True). I don't have time to dig in much further right now but this is odd. Other variables (e.g. v_1206) do have data in them, so I'm not sure where it's getting masked out.