Fix "same" padding torch issue

[sachinprasadhs]

This fixes the issue for all torch layers when used with padding="same" and data_format="channels_first".

This was not caught in any of the test case due to the reason that it only happens with the above condition and when the padding_size is obtained from _compute_padding_length is something like ((0,0), (1, 1))

Fixes: #20235

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 20.00000% with 4 lines in your changes missing coverage. Please review.

Project coverage is 75.78%. Comparing base (7628a18) to head (b5e23e2).
Report is 208 commits behind head on master.

Files with missing lines	Patch %	Lines
keras/src/backend/torch/nn.py	20.00%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #20270      +/-   ##
==========================================
+ Coverage   66.34%   75.78%   +9.44%     
==========================================
  Files         508      527      +19     
  Lines       48264    60532   +12268     
  Branches     8882    12130    +3248     
==========================================
+ Hits        32020    45874   +13854     
+ Misses      14425    12609    -1816     
- Partials     1819     2049     +230     

Flag	Coverage Δ
keras	75.68% <20.00%> (+9.34%)	⬆️
keras-jax	64.27% <20.00%> (?)
keras-numpy	59.80% <20.00%> (?)
keras-tensorflow	65.49% <20.00%> (-0.85%)	⬇️
keras.applications	?
keras.applications-jax	?
keras.applications-numpy	?
keras.applications-tensorflow	?
keras.applications-torch	?

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]

Thanks, Sachin. It looks like relevant torch tests are failing, can you take a look? https://github.com/keras-team/keras/actions/runs/10931265787/job/30345878233?pr=20270

[fchollet]

@sachinprasadhs can you please take a look at the torch test failure?