Numba-mpi

- added wrapper for `bcast` (thanks Delcior)

fix `api` subpackage inclusion in `setup.py` (and add a CI check if package wheel is usable before uploading to pypi)

refactor splitting the codebase into multiple files (thanks Delcior)

API changes:
- returning underlying MPI lib return code from `send`, `recv` and `allreduce` instead of asserting it is zero within these functions (opens up possibilities to use these function in a multi-threaded environmet, and makes the `numba-mpi` a more lightweight wrapper)
- `allreduce` no longer allocates memory for the result, but rather expects it to be passed as argument (type and size not checked!)

thanks Delcior!

Open MPI and Intel MPI added to CI.
Bugfix: MPI_Allreduce() works on Open MPI (issue caused by difference in MPI_Op datatype across different MPI implementations)

Get MPI lib path with `ctypes` `find_library`
Check for Intel MPI dll name (thanks DManowitz for reporting the issue)