Q1_0 ARM repack kernels give a 1.7B model 3.5x prompt processing on a phone.

PR #23492 adds 4x4 and 4x8 NEON repack kernels targeted at Bonsai models. On a Snapdragon 7 Gen 3 with P cores only, Bonsai-1.7B goes from 27.17 to 102.12 t/s prompt processing (+276%) on the NEON+DP 4x4 path and 121.87 t/s (+349%) on NEON+DP+I8MM 4x8, though the wider path is slightly slower at generation. The author published KL-divergence and perplexity tables alongside the speed numbers, mean PPL ratio 1.0055 and 99.2% same-top-p, which is the quality evidence most quantization PRs skip.