remove sorting of bool values since torch 2.4 does not support it on cuda (flagos-ai#731)

Author: iclementine

benchmark/test_special_perf.py

@@ -128,7 +128,7 @@ def sort_input_fn(shape, dtype, device):
         input_fn=sort_input_fn,
         op_name="sort",
         torch_op=torch.sort,
-        dtypes=BOOL_DTYPES + INT_DTYPES + FLOAT_DTYPES,
+        dtypes=INT_DTYPES + FLOAT_DTYPES,
     )
     bench.run()