Add inverse chat templating

[Rocketknight1]

Very experimental PR for now! This PR adds inverse chat templating, where we convert a formatted chat back into message dicts in a universal format. This has been requested several times, and it's critical to allow seamless tool use in pipelines, without requiring users to manually write parsers for each model.

Inverse templating requires the ability to extract text from a large input string. After testing several templates, I found that we can generally handle this by exposing two functions in the inverse templating environment: a slightly modified re.finditer() and a totally unmodified json.loads()

TODO:

• Make some model PRs to test with this, but don't merge yet! (PRs not open, but templates written)
• Add tests for loading/saving of inverse templates
• Add tests for inverse template function
• Test recovery of tools as well as messages
• Make sure I don't need any extra functions
• Refactor chat template tests out of tokenization_common so they're not run for every model
• Add chat template tests to CircleCI
• Make sure extraction works correctly with generation, and add some tests that use (static!) generation outputs
• Add tool use to pipelines (will put this in a separate PR)

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[LysandreJik]

Thanks @Rocketknight1! I'll review ASAP