Add TypeVar variance, bounds, and constraints to type information

[knutwannheden]

Summary

This PR extends the type information system to include detailed metadata about TypeVars, including their variance, upper bounds, and constraints. This allows clients to better understand the properties and restrictions of type variables in Python code.

Key Changes

• TypeVar Variance Support: Added variance field to TypeVar descriptors, supporting covariant, contravariant, and invariant variance modes
• Upper Bound Tracking: TypeVars with upper bounds now expose the bound type through an upperBound field
• Constraints Support: TypeVars with constraints now expose the list of constraint types through a constraints field
• Protocol Updates: Extended TypeDescriptor::TypeVar variant to include optional variance, upper_bound, and constraints fields
• Registry Implementation: Updated type registry to extract and register TypeVar metadata from the semantic database

Implementation Details

• Variance is always included in the response (never skipped during serialization)
• Upper bounds and constraints are optional and only included when present
• The constraints field is an empty vector when not applicable (skipped during serialization)
• All new fields are properly integrated with the existing type registration system

Testing

Added comprehensive integration tests covering:

• TypeVar with explicit covariance
• PEP 695 generic syntax with inferred variance
• TypeVar with upper bounds
• TypeVar with constraints
• Unbounded/unconstrained TypeVars (ensuring fields are absent)