Add support for GPT-4-O, "Omni" model

[pkallos]

@dqbd seems the encoding model has changed, need to build/test support with o200k_base

[pkallos]

OK, I have made the following changes in order to support the new encoding model:

• Copied the ranks source url, special token callouts, and PAT_STR regular expressions from the openai/tiktoken project
• Added supports Rust code for constructors on the o200k_base encoder
• Adjusted the regex-rewriter in scripts/ranks.ts to support global replace of ?i: patterns
• Added and adjusted a unit test for the new encoding in wasm/test/test_simple_public.test.ts which appears to work
• Pointed the gpt-4o model_to_encoding mappings to refer to the new o200k_base encoding

This has been tested as follows:

1. build the project with turbo build
2. run the tests with turbo test

[Pckool]

Please merge this asap @dqbd 🙏 breaks gpt4o support on langchain

[dqbd]

Thank you so much @pkallos!

[gonzalonunez]

Amazing! Can we draft a new release?

[elliotmassen]

Hey @dqbd, do you know when version 1.0.15 might be published to NPM?