qview Find tool wrongly prioritizes camera model over map projection

[rvwagner]

ISIS version(s) affected: 8.0.3, also every prior version since at least 3.4.10

Description

Currently in qview, if you have a map-projected image with an attached camera model, when you use the Find tool in "Match Scale" mode, the zoom level is calculated based on the camera model, not the map projection. This leads to the image zooming in and out slightly, apparently at random (actually correlated with slant distance, I think), if you click around in another image. This zooming causes the two images to not actually have matching scales, because any slant-distance-based scaling has already been compensated for during map-projection, which messes with using the blink comparison tool.

Correct behavior would be for the Find tool to ignore the camera model if an image is map projected, and rely entirely on the projection information.

How to reproduce

wget https://pds.lroc.asu.edu/data/LRO-L-LROC-2-EDR-V1.0/LROLRC_0023/DATA/ESM2/2015113/NAC/M1184418087RE.IMG
lronac2isis from=M1184418087RE.IMG to=M1184418087R.cub
spiceinit from=M1184418087R.cub web=true
cam2map from=M1184418087R.cub to=M1184418087R.map.cub

# Remove spice data
cp M1184418087R.map.cub M1184418087R.nospice.cub
editlab from=M1184418087R.nospice.cub grpn=Kernels op=delg

# could use automos to get the same result:
#ls M1184418087R.map.cub > M1184418087R.lis
#automos from=M1184418087R.lis mos=M1184418087R.nospice.cub

qview M1184418087R.map.cub M1184418087R.nospice.cub

Zoom to 1:1, link the images, and use the Find tool to click around in M1184418087R.nospice.cub. Note that despite these two files being literally the same image with the same map projection, the zoom level on M1184418087R.map.cub varies slightly, breaking the correct scale matching (for the NAC, these errors tend to be small; for ShadowCam, the scale error can get much more extreme; I'm not sure what the cause is). You can see in the blink tool that M1184418087R.map.cub loses a pixel or two, again despite these being literally pixel-for-pixel identical images.