genereate_latlon_boundary does not work for an approach/limb image

[kberryUSGS]

generate_latlon_boundary doesn't seem to work for an approach/limb image, as it appears to only project points on the boundary of the image. If these pixels don't have the target in them, I'm not sure that the result will make sense.

[jessemapel]

I wonder if there is an existing algorithm for handling this type of stuff.

[scsides]

If I remember correctly, the image where this was discovered did return a polygon even though none of the image edge pixels intersected the target. This should be looked at.

[jessemapel]

@scsides This is probably due to how imageToGround behaves in USGSCSM when it doesn't intersect the body. It will return the point on the image ray closest to the body.

[scsides]

It would be nice to find an existing library for this, but the algorithm in ISIS does produce a valid polygon for most cases including full image, full disk, limb, north pole, south pole, longitude boundary, but not both north and south poles in the same image.

[scsides]

That's concerning. Is there anything that tells us it failed, but returned an invalid result anyway?

[scsides]

@jessemapel says there should be a warning added if this occurs, and that we should check to make sure this is happening.