quant: throw a nice error message for allclose with quantized inputs (p…

…ytorch#49802)

Summary:
Pull Request resolved: pytorch#49802

Currently `torch.allclose` is not supported with quantized inputs.
Throw a nice error message instead of a cryptic one.

Test Plan:
```
torch.allclose(x_fp32, y_fp32)

torch.allclose(x_int8, y_int8)
```

Imported from OSS

Reviewed By: supriyar

Differential Revision: D25693538

fbshipit-source-id: 8958628433adfca3ae6ce215f3e3ec3c5e29994c

aten/src/ATen/native/TensorCompare.cpp

@@ -38,6 +38,8 @@ Tensor isclose(const Tensor& self, const Tensor& other, double rtol, double atol
   TORCH_CHECK(self.scalar_type() == other.scalar_type(), self.scalar_type(), " did not match ", other.scalar_type());
   TORCH_CHECK(!(self.is_complex() && equal_nan),
     "isclose with equal_nan=True is not supported for complex inputs.");
+  TORCH_CHECK(!(self.is_quantized() || other.is_quantized()),
+    "isclose is not supported for quantized inputs.");
 
   // Checks that rtol and atol are non-negative
   // Note: consistent with Python's isclose but divergent from NumPy's, which