QGS updated to 1.0.0

[kysolvik]

QGS package was recently updated to 1.0.0. Awesome that it's still being developed, but it's breaking tests currently.

Pinned it at 0.2.8 for now, but need to dive into the new documentation to update our code to work with 1.0.0.

The error:

ValueError: This operation requires zero fill values, but argument 0 had a fill value of -0.0.

[kysolvik]

Issue may actually be sparse 0.15.4 -> 0.16.0. Fixed with ea9d8f6 for now, but needs more investigation. Somehow, QGS is generating fill values of -0.0, while sparse requires fills of 0 for matrix multiplication. Could just be a floating point arithmetic error?