Add CUDA convolution example with profiling support

[Alessandro624]

Description

This PR introduces a new CUDA convolution module, providing a practical example of convolution kernels along with a complete build, execution, and profiling workflow.

Key changes

• Added a new convolution/ module including:

  • CUDA convolution kernels implementing the core computation.
  • Makefile to standardize compilation.
  • run.sh for straightforward execution.
  • profile_nvprof.sh to collect GPU performance metrics.
  • README documenting usage, configuration, and profiling steps.

Impact

The convolution example expands the set of GPU workloads available in the repository, complementing matrix multiplication with a computation pattern widely used in signal processing and deep learning. It provides a solid baseline for performance analysis and kernel optimization experiments.