Batched FFT buggy

[geggo]

This issue is to remind me that the code for batched FFTs is still buggy.
clFFT supports only batching along a single non-transform axis, but in some cases it is possible to collapse several non-transform axes.

gpyfft might fail to properly collapse

• if array contains dimensions of size 1: strides are non-increasing, sorting is ambiguous
• batched 2d transforms: depending on memory layout, transformed array might get transposed