fix: Resolve fixed-size array issue

Replace .ToArray() with .ToList() to allow dynamic modification of network_nodes in MapGraphNetwork()

Replaced .ToArray() with .ToList() to resolve the issue where .Add() was called on a fixed-size array.

This preventing the "Collection was of a fixed size" error when called something like this var model = keras.Model(new Tensors(new Tensor[] { encoder_inputs, decoder_inputs }), outputs: decoder_dense);

fix: Support for training a multi-input model using a dataset.

fix: More generic array cast

test: more gradient optimizer tests

fix: regularizer serialization problem

fix: Avoid modifying augmented graph view collection upon traversal

refactor: python test common class moved to a separate common assembly

Fix Conv2dTranspose bias

test: Gradient descent  optimizer tests

passing

fix: fix the validation_pack when multiple input

refactor: refacter the place of the kernel unittest folder