metal : add CONV_2D_DW (depthwise convolution) support (#21565)

* metal : add CONV_2D_DW (depthwise 2D convolution) support

* test : add perf cases for CONV_2D_DW

* metal : use 3D dispatch for CONV_2D_DW kernel

* metal : add channel-tiled CONV_2D_DW kernel for non-contiguous layouts

* metal : simplify CONV_2D_DW dispatch and trim comments

* metal : merge duplicate CONV_2D_DW pipeline getters

* tests : add F16 CONV2D_DW tests

* cpu : fix F16 kernel support for CONV_2D_DW

* tests : remove commented-out CONV_2D_DW test block

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

ggml/src/ggml-metal/ggml-metal-impl.h

@@ -656,6 +656,34 @@ typedef struct {
     int32_t  d1;
 } ggml_metal_kargs_conv_2d;
 
+typedef struct {
+    uint64_t nb00;  // kernel strides
+    uint64_t nb01;
+    uint64_t nb02;
+    uint64_t nb10;  // input strides
+    uint64_t nb11;
+    uint64_t nb12;
+    uint64_t nb13;
+    uint64_t nb0;   // output strides
+    uint64_t nb1;
+    uint64_t nb2;
+    uint64_t nb3;
+    int32_t  IW;    // input width
+    int32_t  IH;    // input height
+    int32_t  KW;    // kernel width
+    int32_t  KH;    // kernel height
+    int32_t  C;     // channels (IC == OC for depthwise)
+    int32_t  OW;    // output width
+    int32_t  OH;    // output height
+    int32_t  N;     // batch size
+    int32_t  s0;    // stride x
+    int32_t  s1;    // stride y
+    int32_t  p0;    // padding x
+    int32_t  p1;    // padding y
+    int32_t  d0;    // dilation x
+    int32_t  d1;    // dilation y
+} ggml_metal_kargs_conv_2d_dw;
+
 typedef struct {
     uint64_t  ofs0;
     uint64_t  ofs1;