[BLD] Bound tokenizers version

[itaismith]

Description of changes

Latest tokenizers package does not have binaries for python 3.8, so bounding to the previous working version to allow CI tests to pass.

Test plan

• Tests pass locally with pytest for python, yarn test for js, cargo test for rust

Documentation Changes

N/A

[github-actions]

Reviewer Checklist

Please leverage this checklist to ensure your code review is thorough before approving

Testing, Bugs, Errors, Logs, Documentation

• Can you think of any use case in which the code does not behave as intended? Have they been tested?
• Can you think of any inputs or external events that could break the code? Is user input validated and safe? Have they been tested?
• If appropriate, are there adequate property based tests?
• If appropriate, are there adequate unit tests?
• Should any logging, debugging, tracing information be added or removed?
• Are error messages user-friendly?
• Have all documentation changes needed been made?
• Have all non-obvious changes been commented?

System Compatibility

• Are there any potential impacts on other parts of the system or backward compatibility?
• Does this change intersect with any items on our roadmap, and if so, is there a plan for fitting them together?

Quality

• Is this code of a unexpectedly high quality (Readability, Modularity, Intuitiveness)

[tazarov]

@itaismith, the tokenizers' team just confirmed that going forward they won't be supporting 3.8 - huggingface/tokenizers#1690 (comment). This is not ideal given that all our CI runs on 3.8 as baseline.

A few figures about 3.8 usage:

• 3.8 is less than 10% of overall downloads (perhaps we can cross-reference with posthog about usage on 3.8)
• 3.8 is EOL (31-Oct-2024) - https://www.anaconda.com/blog/python-3-8-reaches-end-of-life
• Using 3.9+ will allow us to move some of the tests to use testcontainers (which doesn't support 3.8)

We can still continue to build wheels for 3.8, but maybe everything else can be bumped to 3.9. I feel that the tokenizers ditching 3.8 (possibly due to EOL) will be a growing trend in the coming months for other upstream deps.

[HammadB]

Closed in favor of #3211