Bundle CPU .so even when building CUDA backend

[iamthebot]

In #14 we fixed the CUDA .so file not being included in CUDA enabled builds. However, one problem is that bitsandbytes relies on both the presence of CUDA libraries and a GPU working w/ that CUDA version to detect which native lib to load.

On a system with CUDA installed but without a GPU this will cause problems because we don't bundle in the non-CUDA .so in the conda package in that case. We encountered during an import of bitsandbytes on a Ray head node.

This modifies the build to always build the libbitsandbytes_cpu.so even during a CUDA enabled build. This enables non-GPU fallback to work correctly.

I've also added two fellow contributors willing to do some maintenance work on this feedstock.

Checklist

• [x] Used a personal 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-webservices[bot]]

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/meta.yaml) and found it was in an excellent condition.

[shaowei-su]

LGTM, please help add @shaowei-su as maintainers if possible, thanks.

[iamthebot]

@xhochy PTAL, thanks!

[iamthebot]

@xhochy comments addressed, PTAL. Would love to get this merged in since it's blocking us.