Generalize `to_pyro_random_module` for other PPLs

Please let me know if there is a better way to go about this 😃

A clear and concise description of what you wish to refactor. Please include the following:

We would like to generalize to_pyro_random_module in order for other PPLs (eg beanmachine) to integrate with gpytorch.

Modules that will be modified

• to_pyro_random_module -> to_random_module
• rename pyro_random_module_cls prefix from _Pyro to _Random
• We could potentially upstream and refactor to reduce duplication (eg https://github.com/facebookresearch/beanmachine/blob/main/src/beanmachine/ppl/experimental/gp/__init__.py#L15) at the cost of gpytorch taking a dependency on beanmachine (similar to Pyro)

Impact on code structure

• largely (all except upstreaming the integration in the last bullet) cosmetic, no change in structure

Impact on speed

• n/a

Will this be a breaking change?

• Yes, both library BW compatibility due to renaming functions as well as binary compatibility due to renaming class names

[Balandat]

My hot take is that people shouldn’t pickle but instead save state dicts.

[Balandat]

That makes sense to me. I think in regards to backwards compatibility, we could just keep the pyro_ versions of the functions as aliases to the newly introduced ones and raise DeprecationWarnings.