Update outdated workaround for `cusparseSpMV`

[jakirkham]

There is a note in the code that says const needed to be dropped when working with cusparseSpMV in CUDA 11

raft/cpp/include/raft/spectral/detail/matrix_wrappers.hpp

Lines 203 to 214 in 7f57ac6

	// y = alpha*A*x + beta*y
	//(Note: removed const-ness of x, because CUDA 11 SpMV
	// descriptor creation works with non-const, and const-casting
	// down is dangerous)
	//
	virtual void mv(value_type alpha,
	value_type* __restrict__ x,
	value_type beta,
	value_type* __restrict__ y,
	sparse_mv_alg_t alg = sparse_mv_alg_t::SPARSE_MV_ALG1,
	bool transpose = false,
	bool symmetric = false) const

Looking at the CUDA 12 docs for cusparseSpMV, it appears everything is marked const

Given this could we now use const here and anywhere else using cusparseSpMV?

Note: This originally came up in issue ( #2741 ). However it is a bit more nuanced than the other cleanup items in that issue. So have broken it out

[aamijar]

Hi @jakirkham, making a note for myself to check this but we are using cusparsespmv here
https://github.com/rapidsai/raft/blob/main/cpp/include/raft/sparse/solver/detail/lanczos.cuh#L1658

Where would I need to add the const qualifer. Would it be for these variables?
https://github.com/rapidsai/raft/blob/main/cpp/include/raft/sparse/solver/detail/lanczos.cuh#L1637-L1640

Update: After looking at the docs you linked, I guess it would be adding const for the alpha and beta values?

I guess in your original #2741 issue that file wasn't an issue? So maybe I don't need to change anything?