downsized data sets to speed up testing

[abigailsnyder]

Currently devtools::test() takes a long time because our included sample data is large. At some point, we could figure out smaller sample data that works for the tests just as well.

[rplzzz]

The length of the test suite is a problem, but it looks like many of our travis timeouts (mostly on OS X, since the travis OS X instances are slow) are being caused by building the vignettes. I wonder if we could mitigate some of this by eliminating the tutorial1 vignette, which seems to be mostly obsolete.

The other question is, can we speed up the vignette processing? Reducing the resolution of the example data might help. If we halve the resolution in longitude, leaving latitude alone, we halve the size of a bunch of matrices, which could dramatically reduce the time required for some of the SVD and matrix multiplies dramatically (as I read it, for an m x n matrix, SVD is O(m^2*n + n^3) -- with n as the long dimension, that's effectively n^3).