Add stardist fluorescence model

[gnodar01]

We have had stardist H&E for a while, which is pretrained for tissue images, this adds the variant of stardist pretrained on flourescence data from here: https://bioimage.io/#/?tags=stardist&id=10.5281%2Fzenodo.6348084

[Andrea-Papaleo]

Running into error while predicting.
Error in running predictions Error: The shape of dict['input'] provided in model.execute(dict) must be [-1,-1,-1,1], but was [1,128,128,3]

[gnodar01]

The fluorescence model takes one channel inputs. You can find the example image at the zenodo repo, but don't download the jpg (since that's a scaled down, 3 channel, preview image). You should download sample_input_0.tif, which is one channel and 32-bit. Since 32-bit images are not yet supported you have to scale it down to 16-bit in fiji via Image -> type -> 16-bit (I can't upload it here since GitHub doesn't support tiff).

UPDATE: I added the example image in data/images.

I did make the mistake of setting 3 for requiredChannels in the model constructor, which I've now fixed, but there is no difference in behavior if passing in the 3 channel image. The same error is issued; at least on predict piximi doesn't stop you from trying to press predict on a 3 channel image. Filed an issue here.