mlx - passing most layers tests

[acsweet]

This addresses #19571

Most tests in keras/src/layers are passing. This included updated and implemented ops functions (especially in random, image, and numpy)

• is the change in layers/preprocessing/image_preprocessing/random_crop.py RandomCrop.get_random_transformation() okay?
  • mlx can't slice arrays with mlx.arrays (incl scalar values)
• mlx.random.py beta() might need adjusting how it checks if shape is broadcastable with alpha and beta

Couple mlx notes, these be will fixed in their next patch release:

• mlx.core.einsum was failing with ellipsis
• gradient updates were failing with some reverse indexing, e.g. using ::-1 (this affected transpose convolutions), it's currently being handled with a reversed index array and mlx.core.take

(Sorry this got a bit large! I wanted to merge back to the mlx branch before continuing)

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0.92379% with 429 lines in your changes missing coverage. Please review.

Please upload report for BASE (mlx@b9f7141). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
keras/src/backend/mlx/image.py	0.00%	234 Missing ⚠️
keras/src/backend/mlx/random.py	0.00%	104 Missing ⚠️
keras/src/backend/mlx/numpy.py	0.00%	52 Missing ⚠️
keras/src/backend/mlx/nn.py	0.00%	18 Missing ⚠️
keras/src/backend/mlx/linalg.py	0.00%	9 Missing ⚠️
keras/src/backend/mlx/math.py	0.00%	5 Missing ⚠️
keras/src/backend/mlx/core.py	0.00%	3 Missing ⚠️
...s/preprocessing/image_preprocessing/random_crop.py	57.14%	2 Missing and 1 partial ⚠️
keras/src/utils/tf_utils.py	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@          Coverage Diff           @@
##             mlx   #20862   +/-   ##
======================================
  Coverage       ?   72.28%           
======================================
  Files          ?      572           
  Lines          ?    55368           
  Branches       ?     8529           
======================================
  Hits           ?    40020           
  Misses         ?    13409           
  Partials       ?     1939           

Flag	Coverage Δ
keras	72.19% <0.92%> (?)
keras-jax	60.72% <0.92%> (?)
keras-numpy	55.74% <0.92%> (?)
keras-openvino	28.19% <0.23%> (?)
keras-tensorflow	61.22% <0.92%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fchollet]

Thank you for the contribution!