[BUG] `MultiIndex.equals` does not match pandas for numerical indexes with unequal dtypes

**Describe the bug**

```python
import cudf

left = cudf.MultiIndex.from_tuples([(1,)])
right = cudf.MultiIndex.from_tuples([(1.0,)])

print(left.equals(right)) # => False

print(left.to_pandas().equals(right.to_pandas())) # => True
```

**Expected behavior**

This should match pandas. Note that `Index.equals` does do dtype casting.