EC improvements and optimizations

Rewrote the following functions in generic x86_64 assembly:

msac_decode_bool_c: 25.4
msac_decode_bool_avx2: 19.7
msac_decode_bool_adapt_c: 29.7
msac_decode_bool_adapt_avx2: 23.2
msac_decode_bool_equi_c: 23.0
msac_decode_bool_equi_avx2: 17.2
msac_decode_bool_prob_c: 24.3
msac_decode_bool_prob_avx2: 18.8