FFTPower in redshift-space produces nan in the output (related to #663)

[amoradinejad]

In version 0.3.15 of nbodykit, when I run FFTPower to compute the redshift-space wedges, muedges (now) range from -1 to 1 and the output for negative mu values is nan. Furthermore, there are some other nan values in the output for positive values of nu, and largest scale Fourier modes. In #663, the bug about negative mu values is reported, but there is no conclusion of how the issue is resolved. Is there a new version of nbodykit in which the bug is fixed?