[Apple Silicon] Add Metal kernel support in sgl-kernel (#23449)

Signed-off-by: Xiaodong Ye <yeahdongcn@gmail.com>

sgl-kernel/python/sgl_kernel/metal.py

@@ -0,0 +1,30 @@
+"""Python entry points for the sgl_kernel Metal extension."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    import mlx.core as mx
+
+_METALLIB_NAME = "sgl_metal_kernels.metallib"
+
+try:
+    from . import _metal
+
+    _metallib_path = Path(_metal.__file__).resolve().parent / _METALLIB_NAME
+    if not _metallib_path.is_file():
+        raise ImportError(
+            f"{_METALLIB_NAME} not found next to sgl_kernel._metal at {_metallib_path}"
+        )
+    _metal.register_library(str(_metallib_path))
+except ImportError as _exc:  # pragma: no cover - import guarded at call time
+    _metal = None
+    _IMPORT_ERROR: Exception | None = _exc
+else:
+    _IMPORT_ERROR = None
+
+# Python wrappers for the compiled `_metal.*` entry points go below. Each
+# wrapper validates input shapes/dtypes and calls `mx.eval` on its operands
+# before invoking the AOT C++ entry point.