SciJava Table outputs create new rows on error

[gselzer]

I was running the BoneJ2 Use Case, and I noticed in one step my SciJava table added a new row. I've narrowed things down to the following reproducible steps:

1. Run the Use Case up to (and including) the Fractal Dimension Command execution.
2. Run the Surface Area Command as specified, but use the gaussian blur output instead. Eventually, you'll get an ImageJ2 popup (at least on Windows 😅 ) that says "Need a binary image"
3. Run the Surface Area Command as specified, now using the right image. The output table now will have a second row, where the final column has garbage in the first row and the correct value in the second. The rest of the second row is nans:

Ideally, the table would not be making this second row. Ideally after the incorrect execution (which I kept doing because I mixed up the gaussed and threshold images a few times) the garbage value wouldn't be entered, or would just be overwritten when it's run correctly the next time.

I'm guessing that this is an issue upstream, but will have to do some digging to figure out where.