Vamp fix varcut cutoff

[marscher]

The variance cutoff yielded bogus results like:

for d in (0.5, 0.6, 0.7, 0.8, 0.9, 0.95, 0.99, 1.0):
    vamp_dists.dim = d
    print('var cutoff:', d, 'dimensions:', vamp_dists.dimension() )

var cutoff: 0.5 dimensions: 159
var cutoff: 0.6 dimensions: 145
var cutoff: 0.7 dimensions: 129
var cutoff: 0.8 dimensions: 109
var cutoff: 0.9 dimensions: 82
var cutoff: 0.95 dimensions: 62
var cutoff: 0.99 dimensions: 37
var cutoff: 1.0 dimensions: 210

also removed references to mock and six.