Added AVX512 intrinsics-based Ungapped SW alignent kernel

[gsc74]

Summary

This PR adds support for an AVX512 intrinsics-based ungapped Smith-Waterman (SW) alignment kernel to MMseqs2, enabling higher performance on supported hardware.

---

Changes

• Implemented ungapped SW alignment kernel using AVX512 intrinsics.
• Added build-time flag -DHAVE_AVX512=1 to enable AVX512 kernel.

---

Performance

• On 2 socket Intel® Xeon® Platinum 8592+ CPU, the AVX512 kernel shows an end-to-end speedup of ~1.25× for easy-search compared to the AVX2 version.

• Benchmarking was performed using the dataset from:

  Steinegger M and Söding J.
  MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets.
  Nature Biotechnology, doi: 10.1038/nbt.3988 (2017).

• The dataset contains 6370 query sequences and ~30M database sequences.

• A batch size of 1 was used for both AVX2 and AVX512 runs to ensure fair comparison.

---

Build Instructions

To build the MMseqs2 binary with AVX512 support:

mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=Release -DHAVE_TESTS=0 -DHAVE_MPI=0 -DHAVE_AVX512=1 ..
make -j