[jit kernel] Support per_token_group_quant_8bit jit kernel (#18905)

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>

python/sglang/jit_kernel/utils.py

@@ -10,7 +10,6 @@ import torch
 if TYPE_CHECKING:
     from tvm_ffi import Module
 
-
 F = TypeVar("F", bound=Callable[..., Any])
 
 
@@ -73,7 +72,9 @@ class CPPArgList(list[str]):
 CPP_DTYPE_MAP = {
     torch.float: "fp32_t",
     torch.float16: "fp16_t",
+    torch.float8_e4m3fn: "fp8_e4m3_t",
     torch.bfloat16: "bf16_t",
+    torch.int8: "int8_t",
 }