@JiyeongHa found a bug in how I was transforming from neuropythy's retinotopic polar angle convention to the standard one. I had been mapping the polar angle such that 0 corresponded to the right horizontal meridian / positive x-axis, and angle increased clockwise, but angle should've increased counter-clockwise. Further investigation led to the realization that I had done this everywhere where I was defining some sort of polar angle or orientation. This is equivalent to a flip over the vertical axes.
Fixing this doesn't affect our results at all, because the errors cancelled themselves out. The only difference is that all stimuli actually have a negative w_a compared to our original definitions. That is, whereas we originally said forward spirals had w_r=w_a and negative spirals had w_r=-w_a, the reverse is true; while we said all annuli have w_r=0,w_a>0, they all have w_r=0,w_a<0 (this doesn't affect the local spatial frequency, but does matter for the phase.
The equations presented in the paper were all correct, so if anyone was writing an implementation from scratch, this will not affect them. However, the implementations in this code were incorrect.
Other changes are to make sure everything still runs (e.g., updating the neuropythy version as discussed in #2).
Full Changelog: v1.1.1...v2.0.0
