[CK] [CK_Tile] Add GroupConv to Kernel Dispatcher (#5168)

## Motivation

This PR adds CK Tile group convolution (forward, backward-data,
backward-weight) support to the kernel dispatcher, matching and unifying
with the existing dispatcher GEMM infrastructure in architecture and
usability. The dispatcher provides a unified kernel dispatch system with
both C++ and Python frontends, and until now only supported GEMM
operations. This PR enables framework integrators to use the same
declarative kernel workflow for convolutions as they do for GEMM:
declare kernels, build a registry JIT, select kernels within the
registry at runtime, and dispatch to GPU. Future PRs will include
runtime kernel selection heuristics for autotuning of kernel parameters
based on (problem, hardware arch).

## Technical Details

Grouped convolution support has been added to the CK Tile Dispatcher
with generated_conv_backend.hpp enabling dispatcher.run(in, wei, out,
problem) for all 6 conv variants (fwd/bwdd/bwdw x 2D/3D), runtime
heuristic kernel selection, and GroupedConvKernelKey with full
ConvConfigBase fields. Python side adds parallel JIT via
registry.build(max_workers) and heuristic registry.select(). Includes 7
C++ and 6 Python examples covering all directions with CPU reference
validation, and shared infrastructure improvements (BaseRegistry CRTP,
structured exceptions). As a sanity check, JIT compile times for a
single kernel remains the same and for multiple kernels there is better
parallelism:
Kernels | 1 worker | 8 workers
1 | 7.7 s | 7.7 s
2 | 15.9 s | 8.2 s
4 | 33.4 s | 9.7 s
6 | 52.3 s | 10.2 s

## Test Plan

145 ephemeral unit tests have been added to test basic functionality.
All 30 examples/integration tests run end-to-end on gfx950 (MI350): 7
C++ conv, 7 C++ GEMM, 6 Python conv, 10 Python GEMM. CPU reference
validation for forward, backward-data, and backward-weight (2D) in both
C++ and Python examples pass.

## Test Result

30 examples pass. Peak performance: 132 TFLOPS (Batch-32 forward 56x56),
53 TFLOPS (pointwise 1x1). CPU reference accuracy: max_abs_diff < 0.002
for all directions (fp16 vs fp32 reference).

## Submission Checklist

- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

---------

Co-authored-by: Yaswanth Raparti <113389104+yraparti@users.noreply.github.com>

dispatcher/codegen/kernel_config_loader.py

@@ -359,8 +359,8 @@ class ConvTraitConfig:
 
 
 @dataclass
-class ConvKernelConfig:
-    """Complete convolution kernel configuration"""
+class GroupedConvKernelConfig:
+    """Complete grouped convolution kernel configuration"""
 
     tile: ConvTileConfig = field(default_factory=ConvTileConfig)
     trait: ConvTraitConfig = field(default_factory=ConvTraitConfig)
@@ -419,7 +419,11 @@ class ConvKernelConfig:
 
     def kernel_name(self) -> str:
         """Generate kernel name from config"""
-        variant_map = {"forward": "fwd", "bwd_data": "bwdd", "bwd_weight": "bwdw"}
+        variant_map = {
+            "forward": "fwd",
+            "bwd_data": "bwd_data",
+            "bwd_weight": "bwd_weight",
+        }
         var_str = variant_map.get(self.variant, self.variant)
 
         name = f"conv_{var_str}_{self.dtype_input}_{self.ndim}d"
@@ -433,11 +437,11 @@ class ConvKernelConfig:
 
 
 @dataclass
-class ConvKernelConfigSet:
+class GroupedConvKernelConfigSet:
     """A set of convolution kernel configurations loaded from JSON"""
 
     name: str = "default"
-    configs: List[ConvKernelConfig] = field(default_factory=list)
+    configs: List[GroupedConvKernelConfig] = field(default_factory=list)
 
     # Tile parameter ranges
     tile_m_values: List[int] = field(default_factory=lambda: [128])
@@ -481,7 +485,7 @@ class ConvKernelConfigSet:
     layout: str = "nhwgc"
     gpu_targets: List[str] = field(default_factory=lambda: ["gfx942"])
 
-    def generate_configs(self) -> Iterator[ConvKernelConfig]:
+    def generate_configs(self) -> Iterator[GroupedConvKernelConfig]:
         """Generate all kernel configurations (cartesian product)"""
         # Tile parameters
         tile_params = itertools.product(
@@ -548,7 +552,7 @@ class ConvKernelConfigSet:
                             double_smem_buffer=trait[6],
                             num_groups_to_merge=trait[7],
                         )
-                        yield ConvKernelConfig(
+                        yield GroupedConvKernelConfig(
                             tile=tile_cfg,
                             trait=trait_cfg,
                             dtype_input=self.dtype_input,
@@ -599,7 +603,9 @@ class ConvKernelConfigSet:
         return tile_count * trait_count * extra_count * len(self.gpu_targets)
 
 
-def load_conv_kernel_configs(json_path: str | Path) -> ConvKernelConfigSet:
+def load_grouped_conv_kernel_configs(
+    json_path: str | Path,
+) -> GroupedConvKernelConfigSet:
     """
     Load convolution kernel configurations from a JSON file.
 
@@ -607,14 +613,14 @@ def load_conv_kernel_configs(json_path: str | Path) -> ConvKernelConfigSet:
         json_path: Path to JSON configuration file
 
     Returns:
-        ConvKernelConfigSet with all parameter values loaded
+        GroupedConvKernelConfigSet with all parameter values loaded
     """
     json_path = Path(json_path)
 
     with open(json_path) as f:
         data = json.load(f)
 
-    config_set = ConvKernelConfigSet()
+    config_set = GroupedConvKernelConfigSet()
 
     # Name
     config_set.name = data.get("kernel_set_name", json_path.stem)
@@ -680,15 +686,15 @@ def load_conv_kernel_configs(json_path: str | Path) -> ConvKernelConfigSet:
 
 
 def generate_cpp_conv_kernel_set_declaration(
-    config_set: ConvKernelConfigSet,
+    config_set: GroupedConvKernelConfigSet,
     set_name: Optional[str] = None,
 ) -> str:
     """
-    Generate C++ DECL_CONV_KERNEL_SET code from a ConvKernelConfigSet.
+    Generate C++ DECL_GROUPED_CONV_KERNEL_SET code from a GroupedConvKernelConfigSet.
     """
     name = set_name or config_set.name
 
-    lines = [f"DECL_CONV_KERNEL_SET({name},"]
+    lines = [f"DECL_GROUPED_CONV_KERNEL_SET({name},"]
 
     for config in config_set.generate_configs():
         line = f'    .add("{config.dtype_input}", "{config.variant}", {config.ndim}, '