Update silica thin film loss

The current slope of 0.4 is not really supported by the data. This was discussed in gwinc/matgwinc#10 and @aaron-buikema, @matthew-evans, @lee-mcculler, and I came to the consensus that the value should be set to zero. This makes the numbers for both the low- and high-index loss angles (and their slopes) consistent with the analysis in Slawek's 2018 PRD.

This was already implemented in gwinc/matgwinc!7 and there is no reason not to implement it in pygwinc also. This addresses #35 (closed).

Additionally I have put the loss angle at 100 Hz back to 4e-5 (it was changed to 5e-5 at some point in the past). This is the value @nicholas.demos and @slawomir.gras now assume.