Re-test with chacha20poly1305 0.9

[str4d]

Hi! I was wondering if you could upgrade to chacha20 0.8 and chacha20poly1305 0.9, and re-run your benchmarks (since you have the machine you ran them on, and it isn't documented in the repo, so I can't reproduce them)? Since chacha20 0.6 we've made several improvements to the ChaCha20 AVX2 backend, added a ChaCha20 NEON backend for aarch64 targets, and added backend autodetection (I believe your current benchmarks only used the software implementation, since they don't appear to enable the avx2 feature which was necessary in chacha20 0.6). I'm interested to see how much of an improvement these make compared to the rest of the field!