Add parallel compile from pybind11

[wiredfool]

Fixes #8971 .

Changes proposed in this pull request:

• Use the ParallelCompile from pybind11 to compile the c extension in parallel.

With:

$ make clean && time make install
...
real	0m9.226s
user	0m33.522s
sys	0m3.639s

$ make clean && time make debug
...
real	0m4.724s
user	0m15.480s
sys	0m2.941s

On main:

$ make clean && time make install
...
real	0m21.224s
user	0m20.892s
sys	0m2.613s

$ make clean && time make debug
...
real	0m10.243s
user	0m9.450s
sys	0m2.123s