Pydantic Plugin: Implement Annotated with Field constraints support

[genro]

Context

Add support for Pydantic's Annotated type with Field constraints in the plugin:

from typing import Annotated
from pydantic import Field

@sw
def create_user(
    name: Annotated[str, Field(min_length=2, max_length=50)],
    age: Annotated[int, Field(ge=0, le=120)],
    email: Annotated[str, Field(pattern=r'^[\w\.-]+@[\w\.-]+\.\w+$')]
):
    ...

Implementation Plan

1. Analyze Current Plugin

• Review PydanticPlugin.on_decore to understand how signature and type hints are extracted
• Identify the exact point where parameters are converted into the fields map for create_model()

2. Handle Annotated Types

Create a helper function _resolve_annotation(annotation) -> tuple[type, FieldInfo|None]:

• Use typing.get_origin / typing.get_args to check if annotation is Annotated
• Return the base type (first argument) and the first FieldInfo object if present
• Support chains like Annotated[int, Field(...), SomeMetadata], ignoring non-Field metadata
• Return (type, None) for non-Annotated types

Integrate the helper in the parameter loop:

• If FieldInfo is present, use it for defaults and constraints instead of simple (hint, default)
• Pass (hint, field_info) tuples to create_model() when available

3. Model Creation

• When calling create_model(), pass tuples of (hint, field_info) when available
• Fall back to (hint, default) or (hint, ...) for parameters without FieldInfo
• Ensure backward compatibility: parameters without hints or non-Annotated continue to work

4. Tests

Add dedicated tests (in tests/plugins/test_pydantic.py or similar):

• Decorate a handler with Annotated[str, Field(min_length=2)] and verify invalid input raises ValidationError
• Cover at least:
  • String constraints (min_length, max_length, pattern)
  • Numeric constraints (ge, le, gt, lt)
  • Mixed: some parameters with Field, some without
• Verify all existing tests continue to pass

5. Documentation

• Update llm-docs/README.md or plugin-specific documentation to mention Annotated support
• Add practical example from this issue (name, age with constraints)
• Document that this provides fine-grained control for validation

Work Breakdown

• 🔧 Implementation: Helper function + integration (~50 lines)
• ✅ Tests: Comprehensive test cases covering constraints (~100 lines)
• 📝 Documentation: Examples and usage guide

Related Issues

• Pydantic Plugin: Add support for Literal and Enum types #15 - Literal and Enum support (complementary feature)
• Pydantic Plugin: Test, document and configure Union type coercion behavior #26 - Union type coercion (can combine with Field(strict=True))
• Supersedes Pydantic Plugin: Add support for Annotated with Field constraints #14 (original request, closed in favor of this one)

Priority

Medium - Natural enhancement of existing Pydantic plugin, no new dependencies required.