CUDA kernel grid and block dimension configuration utilities.

Provides helper classes for configuring CUDA kernel launch parameters with automatic grid dimension calculation, overflow checking, and validation. These classes simplify the process of determining optimal block and grid sizes for 1D and 2D computational problems.

The Grid classes automatically calculate the number of blocks needed to cover all elements, handle small problem sizes gracefully, and provide safety guarantees through input validation and overflow detection. Grid2D is optimized for column-major matrix operations with intelligent block dimension selection based on problem size.

Classes:

culite::GridXD : Base class for CUDA kernel grid configuration
culite::Grid1D : 1D CUDA kernel grid configuration with automatic block calculation
culite::Grid2D : 2D CUDA kernel grid configuration with adaptive block sizing