Maybe a small bug about RDN

[ryanxingql]

Code: https://github.com/open-mmlab/mmediting/blob/e1d488323096c0850a08919ffdb03e0139ca66d7/mmedit/models/backbones/sr_backbones/rdn.py#L54

The output channel number of self.lff is channel_growth.
However, the input channel number of x is in_channels.

In the forward function,
https://github.com/open-mmlab/mmediting/blob/e1d488323096c0850a08919ffdb03e0139ca66d7/mmedit/models/backbones/sr_backbones/rdn.py#L66
the residual learning may lead to error because of the misalignment.

I refer to the original code of RDN. The output channel number should be the same as the input channel number.

[zengyh1900]

Please check this issue. @plyfager

[ryanxingql]

I have raised a PR: #1292

[plyfager]

Thanks for your contribution, #1292 has been merged.