CxxUtils::vec allow const source (src) argument for vload/vstore

vload(VEC& dst, vec_type_t<VEC>* mem_addr)

--->

vload(VEC& dst, vec_type_t<VEC> const* src)

similar for vstore follows the memcpy conventions.

2. I also updated comments

3. I moved the using mask_type_t = typename mask_type<VEC>::type; machinery just below the using vec_type_t = typename vec_type<VEC>::type; as seems to fit better there (have these one after the other)