Unify the data pointer in the AVX kernel interface

Issue/feature addressed

Previously there are too many kinds of data pointers:

simd_type *
simd_type::scalarType *
simd_type::vectorType *

And also too many different workspaces

std::vector<simd_t, tinysimd::allocator<simd_t>>
alignas(simd_t::alignment) TData[nqTot*simd_t::wdith];
static_cast<TData *>(::operator new[](nqTot *simd_t::width * sizeof(TData), std::align_val_t(simd_t::alignment)));

Proposed solution

Leave only 1 pointer simd_t * . We can reinterpret TData * ** to simd_t * freely and vice versa;**

Leave only std::vector<simd_t, tinysimd::allocator<simd_t>> as a workspace;

Technically, using a plain array when size is a compile-time constant should guarantee the memory allocated on stack rather than heap, which is more efficient.

But so far, no clear evidence shows this has a significant impact.

Implementation

Tests

Suggested reviewers

@CFD-Xing

Notes

Please add any other information that could be useful for reviewers.

Checklist

• Functions and classes, or changes to them, are documented.
• [ ] User guide/documentation is updated.
• [ ] Changelog is updated.
• Suitable tests added for new functionality.
• Contributed code is correctly formatted. (See the contributing guidelines).
• License added to any new files.
• No extraneous files have been added (e.g. compiler output or test data files).