feat: customly overload the reducers that call cuda-compute kernels

[maxymnaumchyk]

The purpose of this PR is to have custom reducers for cuda-compute (cccl) kernels (in a similar manner to JAX backend).

[codecov]

Codecov Report

❌ Patch coverage is 76.71233% with 51 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.28%. Comparing base (441d5bd) to head (f567021).
⚠️ Report is 1 commits behind head on main.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
src/awkward/_connect/cuda/reducers.py	81.00%	38 Missing ⚠️
src/awkward/_connect/cuda/_compute.py	7.14%	13 Missing ⚠️

❌ Your patch status has failed because the patch coverage (76.71%) is below the target coverage (98.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

Files with missing lines	Coverage Δ
src/awkward/_backends/cupy.py	98.43% <100.00%> (ø)
src/awkward/_connect/cuda/__init__.py	92.92% <100.00%> (+0.07%)	⬆️
src/awkward/_connect/cuda/_compute.py	57.24% <7.14%> (-2.61%)	⬇️
src/awkward/_connect/cuda/reducers.py	81.00% <81.00%> (ø)

... and 7 files with indirect coverage changes

[github-actions]

The documentation preview is ready to be viewed at http://preview.awkward-array.org.s3-website.us-east-1.amazonaws.com/PR3894

[ianna]

@maxymnaumchyk - Great! Please check my comments. Thanks.

[ianna]

@maxymnaumchyk - great! Thanks. This one should go first and the #3840 one next.