Simplify sgr_x_by_x calculations

                        old   ->    new

sgr_3x3_8bpc_c:      196497.7    186258.0
sgr_3x3_8bpc_ssse3:   81948.3     78151.6
sgr_3x3_8bpc_avx2:    35997.4     35478.9
sgr_5x5_8bpc_c:      202810.7    197039.0
sgr_5x5_8bpc_ssse3:   58445.4     57199.3
sgr_5x5_8bpc_avx2:    25964.5     25464.5
sgr_mix_8bpc_c:      383551.3    370946.5
sgr_mix_8bpc_ssse3:  129523.3    121436.6
sgr_mix_8bpc_avx2:    56125.1     54738.1

sgr_3x3_10bpc_c:     204908.4    198493.6
sgr_3x3_10bpc_ssse3:  94347.5     90350.1
sgr_3x3_10bpc_avx2:   37039.5     36606.2
sgr_5x5_10bpc_c:     195876.9    191415.5
sgr_5x5_10bpc_ssse3:  62818.1     59975.1
sgr_5x5_10bpc_avx2:   27420.0     26813.8
sgr_mix_10bpc_c:     395164.6    373924.9
sgr_mix_10bpc_ssse3: 149043.0    140697.1
sgr_mix_10bpc_avx2:   62856.1     60661.7