This patch implementation is primarily focused on
- Lowering to LLVM IR, by generating appropriate kmpc_alloc() and kmpc_alligned_alloc() calls for variable(s) and before the end of scope generating kmpc_free() for the same variable(s).
- Also handled, usage of array variables in the OMP ALLOCATE directive.
- Define omp.allocate_free operation in MLIR and slight changes to existing MLIR definition of ALLOCATOR clause.
- Add test cases for variations of usage of OMP ALLOCATE directive and its clauses ALIGN and ALLOCATOR.