x86: Improve SSSE3 SGR asm

Checkasm numbers:

Zen 4                     Old        New
sgr_5x5_8bpc_ssse3:     31509.5    30452.0
sgr_3x3_8bpc_ssse3:     44256.5    41791.9
sgr_mix_8bpc_ssse3:     70337.2    66383.8
sgr_5x5_10bpc_ssse3:    33681.4    31481.8
sgr_3x3_10bpc_ssse3:    48955.3    45495.9
sgr_mix_10bpc_ssse3:    79951.3    73002.1

Rocket Lake
sgr_5x5_8bpc_ssse3:     61661.7    58291.5
sgr_3x3_8bpc_ssse3:     88814.0    78772.7
sgr_mix_8bpc_ssse3:    138456.1   122995.2
sgr_5x5_10bpc_ssse3:    66229.1    60628.0
sgr_3x3_10bpc_ssse3:   100219.0    85590.7
sgr_mix_10bpc_ssse3:   157864.2   137124.9