torchdrug.patch.py breaks use of ConcatDataset

[bbrier]

Creating a torch ConcatDataset fails after running torchdrug.patch.py, so by extension every torchdrug import.

Minimal example:

from torch.utils.data import Dataset, ConcatDataset
import torchdrug

class DummyDataset(Dataset):
    def __init__(self, size=50):
        self.size=size
        
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        return idx

dummy = ConcatDataset([DummyDataset(), DummyDataset()]) 

The issue appears to be caused by the following code snippet from patch.py

Dataset = dataset.Dataset
for name, cls in inspect.getmembers(dataset):
    if inspect.isclass(cls) and issubclass(cls, Dataset):
        cls = core.make_configurable(cls)
        cls = R.register("dataset.%s" % name)(cls)
        patch(dataset, name, cls)
importlib.reload(torch.utils.data)

Expected behavior: ConcatDataset is correctly initialised
Actual behavior: Initialisation of ConcatDataset raises {TypeError}__init__() missing 1 required positional argument: 'datasets'

python version: 3.9.16
torch version: 1.13.1+cu117
torchdrug version: 0.2.1

[toooooodo]

I'm encountering the same issue.