Updated conv function to use a NumPy-based implementation

[shashaka]

I have updated the conv function in nn.py to use a NumPy-based implementation, referring to the one in conv_test. If there are any concerns regarding performance, please feel free to share them so I can close this PR if needed.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 92.80576% with 10 lines in your changes missing coverage. Please review.

Project coverage is 82.49%. Comparing base (dac30fa) to head (3105578).

Files with missing lines	Patch %	Lines
keras/src/backend/numpy/nn.py	92.80%	7 Missing and 3 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #21051      +/-   ##
==========================================
+ Coverage   82.47%   82.49%   +0.02%     
==========================================
  Files         563      563              
  Lines       53837    53974     +137     
  Branches     8359     8380      +21     
==========================================
+ Hits        44402    44526     +124     
- Misses       7394     7403       +9     
- Partials     2041     2045       +4     

Flag	Coverage Δ
keras	82.31% <92.80%> (+0.02%)	⬆️
keras-jax	63.63% <0.00%> (-0.17%)	⬇️
keras-numpy	58.83% <92.80%> (+0.07%)	⬆️
keras-openvino	32.63% <0.00%> (-0.09%)	⬇️
keras-tensorflow	63.95% <0.00%> (-0.17%)	⬇️
keras-torch	63.66% <0.00%> (-0.17%)	⬇️

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.

🚀 New features to boost your workflow:

• ❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.