This follows up on the fix from #173, which inadvertantly introduced a global phase on Rx/Ry rotations by 0.0 (and odd increments of Pi). This avoids the problem by checking the other part of the matrix to determine if the phase is required rather than applying it unconditionally.
This follows up on the fix from #173, which inadvertantly introduced a global phase on Rx/Ry rotations by 0.0 (and odd increments of Pi). This avoids the problem by checking the other part of the matrix to determine if the phase is required rather than applying it unconditionally.