Create seqscore lib and migrate to Pydantic

[Wollaston]

This PR is twofold:

• Convert the existing seqscore repo into a workspace with the main seqscore cli as the workspace root, and the new seqscore-lib project as the library backend for the CLI. The intention is so that this can be published/distributed separately and used as a library. This uses uv to effectively manage and package this.
• Use pydantic instead of attrs

Note Pydantic provides automatic type validations and conversions. So, depending on configuration, some LSPs/Type Checkers may not know this and throw errors. If you run mypy and the pre-commit script, it will show that all types have been statically checked.

One notable change is with the LabeledSequence implementation. Previously, this overwrote the __iter__ dunder method to return self.tokens as an Iterable. This couldn't be done with Pydantic as it relies on __iter__ for its internal validation. There only appeared to be a few calls that expected this functionality, so those just call the tokens field of the instance to get an Iterable and the Pydantic default __iter__ is maintained for its expected internal use elsewhere.

Pydantic does not currently support opting certain fields out of __hash__ and __eq__ so for LabeledSequence again, those are manually implemented to maintain the original functionality.

[ConstantineLignos]

I took a quick look and things look sane. Let's meet next week to talk over the restructuring of the library vs cli. I want to get an idea of if it's absolutely necessary and what alternatives are to the current approach.

[Wollaston]

Closed in favor of not restructuring library at this time