[air][example] Dreambooth fine-tuning example fixes

[justinvyu]

Why are these changes needed?

This examples makes a few improvements on the dreambooth fine-tuning example:

• Fixes a bug with the batching logic (batch size > 1 used to be buggy). See below for some samples before/after the fix, training with batch size = 2.

Before fix (this is some random dog)	After fix (this is the subject dog)

• Simplifies the data preprocessing logic (the "image" vs "image_1" columns from a zipped dataset was confusing before)
• Switches to the dog dataset, and fixes the lego car dataset class to just car (from @woshiyyya's experiments)
• Updates the doc guide a bit, so that it can be linked in an upcoming workspace template. (The template will only provide some basic instructions on how to run.)

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 added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• 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 :(

[woshiyyya]

Thank you Justin! LGTM. Added a few comments.

Maybe in workspace template, we can also add this online inference notebook to let user play with their model interactively with customized prompts. https://github.com/woshiyyya/dreambooth-demo/blob/master/serve.ipynb

[justinvyu]

Release test passed here (see Artifacts for a sample generated image): https://buildkite.com/ray-project/release-tests-pr/builds/46271#01898703-84dd-4596-8240-89a106924b04

Rendered doc updates: https://anyscale-ray--37504.com.readthedocs.build/en/37504/ray-air/examples/dreambooth_finetuning.html