MLXFast on Linux (CPU)

[Joannis]

Proposed changes

Fills the gap of MLXNN and MLXFast on Linux (CPU mode)

Checklist

Put an x in the boxes that apply.

• I have read the CONTRIBUTING document
• I have run pre-commit run --all-files to format my code / installed pre-commit prior to committing changes
• I have added tests that prove my fix is effective or that my feature works
• I have updated the necessary documentation (if needed)

[davidkoski]

Do both of these files build on all platforms? I think maybe they don't on CPU only linux?

Just thinking out loud here:

• should we use something like #if canImport(Metal) to select which file we use?
• if both build, I wonder if we should switch on stream to route to the correct backend? Then a GPU use could run the CPU implementation if they wanted.

[davidkoski]

If this works then I think this comment could just indicate which case it was (remove the part about the warning).

We have a test for it:

    func testOptionInfos() {
        class Layer1: Module {
            @ParameterInfo var x: MLXArray?
        }