nompi build can use MPI or no-MPI variants at runtime

[xylar]

Checklist

• [x] Used a fork of the feedstock to propose changes
• [x] Bumped the build number (if the version is unchanged)
• [ ] Reset the build number to 0 (if the version changed)
• [x] Re-rendered with the latest conda-smithy (Use the phrase code>@<space/conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• [x] Ensured the license file is being packaged.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe:

• License is not an SPDX identifier (or a custom LicenseRef) nor an SPDX license expression.

[xylar]

In some of my testing where I want parallel versions of libnetcdf and hdf5 but I don't need the parallel version of netcdf4, I'm running into trouble when trying to run python code without mpirun even when it is ostensibly serial code. These changes should allow me to explicitly say that I want the MPI versions of libnetcdf and hdf5 without having to also pick up the MPI version of netcdf4, e.g.:

conda create -y -n test python=3.7 "libnetccf=*=mpi_mpich_*" "hdf5=*=mpi_mpich_*" \
   "netcdf=*=nompi_*"

[xylar]

@conda-forge-admin, please rerender

[xylar]

I will do some local testing of these changes. @conda-forge/netcdf4, any objections to them? If not, I'll merge as soon as I'm done with my local testing.

[xylar]

This worked great in my local tests. I was able to create a conda environment with mpich variants of libnetcdf, etc. but the nompi version of netcdf4 and the difficulties I'd been having with netcdf4 went away!

@conda-forge/netcdf4, any objections to me merging this? If I don't hear anything by this time tomorrow, I'm going to go ahead.

[ocefpaf]

Thanks for fixing this @xylar!