Refactor schemas

[korikuzma]

Some initial tasks:

• Some fields are never used, so we should remove these
• Some fields can be inherited from another class (but this requires renaming of fields to a more generalized name rather than specific to the class itself) #429
• Some fields like so_id and molecule_context are in Tokenizer schemas, but should be moved to Classification schemas. For molecule_context we should use ga4gh.vrsatile.pydantic #428
• Add other token types to token schemas
• Add types to schemas that only use defaults

[github-actions]

Closed by #494.