Refactor inputs using pydantic

[YuchengZhou821]

This pull request refactors the sensor input system to introduce strongly-typed, Pydantic-based configuration classes for each sensor, improving type safety, validation, and code clarity. The changes generalize the Sensor and FuserInput base classes to support specific configuration types, and update all input plugins to use these new config classes. The load_input function is also updated to instantiate sensors with their configuration objects.

Core infrastructure and typing improvements:

• Refactored SensorConfig to inherit from Pydantic's BaseModel, enabling validation and type hints for sensor configuration. The base Sensor and FuserInput classes are now generic over their configuration type, allowing each sensor to specify its own config schema. [1] [2] [3] [4]
• Updated the load_input function in src/inputs/__init__.py to accept a configuration dictionary, dynamically detect and instantiate the appropriate config class for each sensor, and pass it to the sensor's constructor. [1] [2] [3]

Plugin-specific configuration classes:

• For each sensor plugin (e.g., battery_turtlebot4, battery_unitree_go2, dimo_tesla, fabric_closest_peer, face_presence_input, gallery_identities_input), introduced a dedicated Pydantic config class with explicit fields and documentation, replacing generic or ad-hoc config handling. Constructors and config usage within each plugin were updated accordingly. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

Generalization and consistency:

• All sensor plugin classes and their constructors were updated to use the new config types, eliminating the previous pattern of defaulting to a generic SensorConfig and using getattr for config values. [1] [2] [3]

These changes make the input system more robust, maintainable, and ready for further extension.