Split up datatype definition modules

[mpharrigan]

Long overdue.

• prior releases included the qualtran.dtype.XXX aliases for QCDTypes (the data type classes and objects used in your quantum programs)
• This PR moves the source of truth to private submodules in this package and makes the qualtran._infra.dtype imports into aliases
• Most people are probably using the qualtran.XXX re-exports anyways, which remain unchanged.

This is just moving code. The only "change" is moving the assert_to_bits test helper is now in the public API.

[gemini-code-assist]

Code Review

This pull request refactors the quantum data type definitions by moving them from a monolithic file in qualtran/_infra/data_types.py to a new modular package structure under qualtran/dtype. The changes include splitting the types into specialized modules (e.g., _int.py, _fxp.py, _uint.py) and adding comprehensive unit tests for each. A potential issue was identified in the new testing.py utility where the reshape operation may fail when handling symbolic data types, as numpy cannot process symbolic dimensions.