metal : optimize sparse FA + clean-up (#29377)
* metal : cache sparse FA indices in shared memory
Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp
* metal : simplify shared memory size calculation
Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp
* pi : update general
* metal : unroll sparse index load
Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp