Temporarily disable NNC-compile by default to enable PyTorch 2.0

[horizon-blue]

Motivation

I hate to do this, but it looks like the compiler upgrade in PyTorch 2.0 somehow broke the pre-existing NNC usage in Bean Machine. Therefore, the algorithms that have NNC enabled by default are no longer working once we upgrade to PyTorch 2.0.

Until we figure out what's going on with the compiler, let's turn off NNC by default so that our tests can pass under PyTorch 2.0.

Changes proposed

• NNC compile are disable by default now
• Mark Bean Machine as compatible with PyTorch 2.0+ in setup.py so we don't block downstream libraries that depend on us from using newer version of PyTorch.

Test Plan

If CI pass then we should be good.

Types of changes

• [ ] Docs change / refactoring / dependency upgrade
• [x] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to change)

Checklist

• [x] My code follows the code style of this project.
• [x] My change requires a change to the documentation.
• [ ] I have updated the documentation accordingly.
• [x] I have read the CONTRIBUTING document.
• [x] I have added tests to cover my changes.
• [ ] All new and existing tests passed.
• [x] The title of my pull request is a short description of the requested changes.

[horizon-blue]

I'm having trouble reproducing the failing test_constrained_positive_reals test on my machine, though I will take a closer look at a later time.