[kernel slimming] Clean many useless sgl-kernel deprecated kernels (#20277)

sgl-kernel/python/sgl_kernel/memory.py

@@ -1,23 +1,6 @@
 import torch
 
 
-def set_kv_buffer_kernel(
-    k_cache: torch.Tensor,
-    v_cache: torch.Tensor,
-    loc: torch.Tensor,
-    k: torch.Tensor,
-    v: torch.Tensor,
-    fallback: bool = False,
-):
-    try:
-        if fallback:
-            raise RuntimeError("Fallback to torch implementation")
-        torch.ops.sgl_kernel.store_kv_cache(k_cache, v_cache, loc, k, v)
-    except RuntimeError:  # ok, fallback to torch implementation
-        k_cache[loc] = k
-        v_cache[loc] = v
-
-
 def weak_ref_tensor(tensor):
     return (
         torch.ops.sgl_kernel.weak_ref_tensor(tensor)