Better conversion from napari Shapes to Rois

[multimeric]

Closes #63.

[multimeric]

Some manual verification would be great here, because although it doesn't throw an error anymore, I want to be sure that it's producing the correct output.

[pr4deepr]

The error doesn't happen but the coordinates are incorrect.
They are in scaled micron space and not in pixels anymore..

It needs to be transformed in x and y by dividing by dx or dy

For example, if I look at an ROI and view the coordinates in napari, I get:

[array([[206.17889806, 132.66785635],
[206.17889806, 163.98616998],
[240.97702431, 163.98616998],
[240.97702431, 132.66785635]])]

~ cropped image shape is about (150, 35, 31)

but, this is scaled in microns. The actual coordinate should be divided by deskew.dy so the coordinates are in pixels, which will match the actual location in the array. The bbox ROI values should actually be:

array([[1421.92343489, 914.95073342],
[1421.92343489, 1130.93910328],
[1661.91051251, 1130.93910328],
[1661.91051251, 914.95073342]])

~ Actual cropped image shape: (150,216,240)

If I run the cropping, I get a really small image as it uses the bounding box coordinates in microns to perform the cropping.

[multimeric]

Thanks, should be fixed now.

[pr4deepr]

Cropping works fine in the plugin.
However, now, if I select an ROI in the plugin:

and run the cropping, its always the first ROI in the list that gets processed. In the earlier
First ROI in the list is highlighted by a red box below and the processed ROI is the green image on the top

[multimeric]

I don't follow. Is the selected shape supposed to impact cropping?

[multimeric]

TODO:

• Modify the ROI selector to list all shapes from all layers, rather than just listing shape layers
• Synchronise the shape selection with the ROI box
• Warn user in the preview box if multiple ROIs are selected

[multimeric]

Splitting the shape selection issue into a separate PR.