feat: Add Pydantic schemas and config builders for type-safe configuration

[GeorgePearse]

Summary

• Adds comprehensive Pydantic schemas for all major components (backbones, necks, heads, models, data, training)
• Introduces visdet.py_configs module with builder functions providing full IDE autocomplete
• Enhances SchemaRegistry with reverse type name lookup for config validation
• Adds SimpleRunner.from_config() method for creating runners from Pydantic configs

Motivation

This provides a type-safe alternative to YAML/Python config files with full IDE support, making configuration errors catchable at development time rather than runtime.

Example Usage

from visdet.py_configs import mask_rcnn_swin_tiny_coco
from visdet import SimpleRunner

# Create config with full IDE autocomplete
cfg = mask_rcnn_swin_tiny_coco(data_root='/data/coco')
cfg.train_cfg.max_epochs = 24  # Modify with autocomplete

# Create runner from Pydantic config
runner = SimpleRunner.from_config(cfg)
runner.train()

Test plan

• Verify schemas validate example configs correctly
• Test SimpleRunner.from_config() with preset configs
• Confirm IDE autocomplete works with py_configs builders

🤖 Generated with Claude Code

[github-actions]

✅ Skylos Scan: No dead code or security issues detected.