[AIR/Docs] Large model inference examples

[Yard1]

Why are these changes needed?

This PR adds two simple examples showing how to do batch inference with Ray Data using Stable Diffusion and GPT-J models.

Note that the datasets used are very small, allowing them to be ran quickly by both the user and CI. Examples with large scale datasets (>=1 TB) are left as a TODO.

DeepSpeed Inference example will be added in a separate PR.

We are using Ray Data instead of BatchPredictor due to #32923.

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[ericl]

Looks good at a high level. Should we add a TODO to add a larger-scale example? This one is quite small.

[Yard1]

@ericl I think it's fine to keep them small - after all, it is that easy. The DeepSpeed example I am working on will be bigger and more advanced, and the same goes for fine-tuning examples that are next in line.

[richardliaw]

all same comments apply across notebooks. nice work!