feat: Add layered architecture visualization

[AlanPonnachan]

Description

This PR introduces keras.visualization.layered_view(), a new built-in visualization feature that renders Keras models as highly customizable 3D isometric block diagrams. This provides an "at a glance" understanding of model architectures, complementing the existing 2D plot_model utility.

Related issue #21421.

Usage Example

import keras
from keras import layers

model = keras.Sequential([
    keras.Input(shape=(32, 32, 3), name="images"),
    layers.Conv2D(32, 3, activation="relu", padding="same"),
    layers.MaxPooling2D(),
    layers.Conv2D(64, 3, activation="relu", padding="same"),
    layers.MaxPooling2D(),
    layers.Flatten(),
    layers.Dense(128, activation="relu"),
    layers.Dense(10, activation="softmax"),
])

# Generate a 3D isometric view
keras.visualization.layered_view(model, to_file="model.png")

Example outputs

Default 3D View with funnels

2D Flat Mode

3D View without Funnels

Adding a Legend

Custom Text

Custom Color Map and Spacing

Contributor Agreement

• I am a human, and not a bot.
• I will be responsible for responding to review comments in a timely manner.
• I will work with the maintainers to push this PR forward until submission.

[gemini-code-assist]

Code Review

This pull request introduces layered_view, a new visualization tool for Keras models that renders architecture diagrams using aggdraw and Pillow. The implementation includes lazy loading for dependencies and comprehensive tests. Feedback focuses on ensuring the visualization accurately represents non-linear model topologies, adhering to Keras API design guidelines regarding argument count and docstring formatting (including missing Arguments and Example sections), improving error message clarity for unbuilt models, and using more robust introspection for the text_callable argument.

[AlanPonnachan]

@hertschuh Could you please take a look at this?