n_jobs > 1 in aggregation

[leliel12]

currently the parallelization provided by joblib only works for the sub-problem part. The aggregation part is fixed with n_job = 1.

https://github.com/leliel12/sc_drv/blob/master/sc_drv/method.py#L256-L277

The idea is to change again n_jobs=n_jobs to uses paralelism. But some test works