chacha/asm/chacha-x86_64.pl: add dedicated path for 128-byte inputs.

The 128-byte vectors are extensively used in chacha20_poly1305_tls_cipher
and dedicated code path is ~30-50% faster on most platforms.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6626)