This changes the kernels to pack full SVE vectors and reduces the
overall complexity of the inner GEMM loop.

- Adds bgemv T based off of sbgemv T kernel
- Adds bgemv N which is slightly alterated to not use Y as an
accumulator due to the output being bf16 which results in loss of
precision
- Enables BGEMM_GEMV_FORWARD to proxy BGEMM to BGEMV with new kernels

This also improves the testing and generic kernel by re-using the BF16
conversion functions.

Built on top of https://github.com/OpenMathLib/OpenBLAS/pull/5357 and derived from https://github.com/OpenMathLib/OpenBLAS/pull/5287

Co-authored-by: Ye Tao <ye.tao@arm.com>

This improves performance for sbgemv_t by up to 100x on NEOVERSEV1.
The geometric mean speedup is ~61x for M=N=[2,512].

Signed-off-by: Ye Tao <ye.tao@arm.com>

This helps to reduce values going missing as we accumulate.

This patch removes the prefetches from cgemm/zgemm which improves the performance similar to sgemm/dgemm did in #3868, this means I'm happy to enable this on any applicable cores.

I also replicated the unrolling the copies from sgemm and dgemm.

These are generally better and, in some cases, include threading which helps in the cores we're targeting here.

This re-spins #3869 with some additional copy unrolling which helps maintain SYRK performance.

After #3868, the SVE kernels represent a pretty good boost.

This re-uses ARMV8SVE as a base and I'm going to incrementally move everything to use ARMV8SVE in additional patches (as well as fix up anything that's not already in ARMV8SVE).

This adds an SVE implementation to sdot/ddot when available, falling back to the previous Advanced SIMD kernel where there's no SVE implementation for the kernel.

All the targets were essentially treating `dot_thunderx2t99.c` as the Advanced SIMD implementation so I've renamed it to better fit with the feature detection.