Same seed doesn't produce same result due to iter_mut_par

[andresliszt]

The seed argument has been exposed to python in #46, however is not working because we are using some parallel stuff at the fds.

Investigate if in fact parallel helps to performance once this is done #5, if it's removed the seed works as expected
@pwoolvett

[pwoolvett]

Low hanging fruit: we can create a function which selects par/no-par given a cargo setting, which compiles the right one.

So we can test using no-par