Non-determinism

[MariusVanDerWijden]

I've found out that the current libfqfft calculates different fft's depending on the amount of threads it is run in. E.g. the following produces two different outputs.

omp_set_num_threads( 4 );
_basic_parallel_radix2_FFT_inner<int> (v1, 5678, 2, 1);
omp_set_num_threads( 8 );
_basic_parallel_radix2_FFT_inner<int> (v2, 5678, 3, 1);
assert(v1 == v2);

The GPU implementation suffers from the same problem.
I think it has something to do with the padding that is applied here https://github.com/MariusVanDerWijden/gpusnark/blob/4bf8292ea79e05b9404f2742e83a587a32a0982e/fft_host.h#L71