Use stream in mul_add if given and allocator in subset_sum (#438)

* mul_add(): use stream if given, add an optional destination array.
subset_sum(): use allocator if given.

* Remove unused import, add whitespace to make flake8 happy

* Update pycuda/gpuarray.py

Co-authored-by: Andreas Klöckner <inform@tiker.net>

* test_subset_sum: also assert if allocator is used

* subset_sum: use given allocator value (even if None) like other functions

---------

Co-authored-by: Andreas Klöckner <inform@tiker.net>