Refactor AIManager to use ModelManager for model handling

This commit introduces a new `ModelManager` class to encapsulate model-related properties and methods, improving code organization and separation of concerns. Key properties such as `IMAGE_SIZE`, `NUM_DETECTIONS`, and others have been moved to `ModelManager`. The `LoadModelAsync` method now accepts an `IMAGE_SIZE` parameter for better validation and detection calculation.

Event subscriptions for `ClassesUpdated` and `ImageSizeUpdated` have been updated to reference `ModelManager`. The `LoadClasses` method has been modified for better class management, and a `Dispose` method has been added to handle resource cleanup. Additionally, the `CalculateNumDetections` method has been relocated to `MathUtil`, and the code has been cleaned up for improved readability.

Refactor AIManager to use ModelManager for better structure

This commit introduces a new `ModelManager` class to encapsulate model-related properties and methods, improving code organization and separation of concerns. Key changes include:

- Migration of model loading, validation, and class management from `AIManager` to `ModelManager`.
- Updates to event handling for model classes and image size notifications.
- Relocation of the `CalculateNumDetections` method to `MathUtil` as a public static method.
- Simplification of the `AIManager` class by removing redundant properties and methods.
- Refinement of sticky aim logic for better target handling.
- Enhanced error handling and logging during model loading.

These changes enhance maintainability, readability, and functionality related to model management and AI processing.

Author: TaylorIsBlue