bump torchao pin

Performance improvements to lowbit quantized linear metal kernels in torchao. See [AO PR#2167](https://github.com/pytorch/ao/pull/2167) for details. The table below summarizes torchchat's decode speed (tokens/second) on Metal backend on M1 Max 64GB after this update | # bits | Llama 3.2-1B | Llama 3.2-3B | Llama 3.1-8B | | --- | --- | --- | --- | | 1 | 179.96 | 87.01 | 51.10 | | 2 | 186.91 | 98.13 | 54.37 | | 3 | 170.62 | 85.69 | 48.05 | | 4 | 175.15 | 89.54 | 50.51 | | 5 | 147.10 | 70.19 | 38.58 | | 6 | 140.51 | 63.62 | 35.48 | | 7 | 131.27 | 64.19 | 32.69 |