Feat: Add MaxPooling2D and AveragePooling2D Layers

[debug-soham]

Summary

Closes: #108 This PR introduces MaxPooling2D and AveragePooling2D layers to the PyDeepFlow library. This is a crucial enhancement for building effective Convolutional Neural Networks (CNNs), allowing for the down-sampling of feature maps to reduce computational complexity and control overfitting.

Additionally, this PR includes fixes for the Multi_Layer_CNN constructor and the introspection module, ensuring that the model.summary() feature works correctly for both ANN and CNN models and that the entire test suite passes.

Changes Included

• New Layers: Implemented MaxPooling2D and AveragePooling2D classes in pydeepflow/model.py with correct forward and backward passes.
• CNN Constructor Fix: Corrected the Multi_Layer_CNN constructor in pydeepflow/model.py by removing the unused activations parameter, resolving a major conflict with the test suite.
• Introspection Fix: Overhauled the ANNIntrospector and CNNIntrospector classes in pydeepflow/introspection.py to correctly identify all layer types and their properties, ensuring model.summary() is accurate.
• Test Suite Updates:
  • Added tests/test_layers.py with unit tests for the new pooling layers.
  • Corrected calls in tests/test_model_summary.py to align with the updated Multi_Layer_CNN constructor.
• Validator Update: Updated pydeepflow/validation.py to recognize 'maxpool' and 'avgpool' as valid layer types.

Verification

All 200+ tests in the suite, including the new tests for pooling layers, now pass successfully.

[ravin-d-27]

Hi @debug-soham . Good work, thank you!