[schur-complement] validate the arguments and report allocation and launch failures

[zjin-lcf]

The examples themselves are now on master, so what is left here is the hardening of the four schur-complement-* versions. Merged master to resolve the conflicts, keeping master's Makefiles (LAUNCHER, ARCH, SM and CUDA_ARCH/HIP_ARCH overridable from the environment).

Summary of the changes:

• Validate <rows>, <nnz per row> and <repeat>. The CSR row pointers and column indices are 32-bit, as in HiOp, so reject non-positive values, a nonzero count that would not fit an int, and row counts for which m + 1 or rounding m up to the block size would overflow. The check lives in reference.h (valid_problem_size) so the four versions cannot drift apart.
• Report a failed host allocation instead of terminating on bad_alloc, and a failed device allocation in the SYCL version instead of dereferencing a null pointer inside the kernel.
• Check the CUDA and HIP kernel launches, so an invalid launch configuration is reported instead of showing up as a verification failure.
• Give the OpenMP kernels 256 threads per team, matching the 32x8 work-group of the CUDA, HIP and SYCL versions, and hoist the SYCL work-group extents into BLOCK_X/BLOCK_Y as in the other versions.
• Fix the two SYCL kernel comments that still described the original HiOp one-work-item-per-row mapping.

Verified with ./main 1024 32 20 (all PASS): CUDA on a Tesla M40, HIP and OpenMP (AOMP) on an MI100, and SYCL on the M40 through the CUDA backend; the SYCL version also compiles with icpx. Invalid arguments (zero, negative, and out-of-range sizes) are rejected with exit status 1.