version_3.1

replace the old uint320 implementation with epi::uint320_t for poly-1305

version_2.1

• optimized the fallback code for uint320 mul.

version_2.0

• uses uint320 for tag generation, less expensive than the previous uint512.
• supports x86_64 and aarch64/arm64.
• might work on x64?.
• x86 not supported.

version_1.0

initial working implementation

• using uint512 in poly1305 tag generation.