x86: Optimize shifts in 8-bit wiener_filter asm

                          old       new

wiener_5tap_8bpc_c:     51691.0
wiener_5tap_8bpc_sse2:  19979.0   19632.2
wiener_5tap_8bpc_ssse3: 15207.2   14675.6
wiener_5tap_8bpc_avx2:   7730.8    7495.3

wiener_7tap_8bpc_c:     51824.2
wiener_7tap_8bpc_sse2:  24522.0   23849.8
wiener_7tap_8bpc_ssse3: 17342.0   16601.5
wiener_7tap_8bpc_avx2:   8955.2    8521.0