feat: Metal kernels for TurboQuant KV cache (turbo3, turbo4) PrismML-Eng#21

Added Metal shader implementations:
- quantize_turbo3_0 / quantize_turbo4_0 (per-block quantization)
- dequantize_turbo3_0 / dequantize_turbo4_0 (type4x4 and type4 variants)
- kernel_set_rows_turbo template (128-element block size)
- Flash attention instantiations for all dk/dv variants

Added TURBO3_0/TURBO4_0 to Metal device SET_ROWS validation.

Builds successfully. Testing with Qwen 3.5 35B-A3B MoE on M5 Max.

Note: Initial version uses simplified quantization (no rotation matrix)
for Metal compatibility. Full rotation requires custom kernel with extra
buffer bindings — tracked for follow-up.

Co-Authored-By: tturney@psyguard.ai
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: TheTom

ggml/src/ggml-metal/ggml-metal-device.m

@@ -1266,6 +1266,8 @@ bool ggml_metal_device_supports_op(ggml_metal_device_t dev, const struct ggml_te
                     case GGML_TYPE_Q5_0:
                     case GGML_TYPE_Q5_1:
                     case GGML_TYPE_IQ4_NL:
+                    case GGML_TYPE_TURBO3_0:
+                    case GGML_TYPE_TURBO4_0:
                         return true;
                     default:
                         return false;