When Meshroom photogrammetry omits images, model transformation script fails to produce output

[orphefs]

Some images may not be used by the Meshroom photogrammetry executable, as they might not contain enough identifiable features. When running the model transformation script, the algorithm will list the computed poses (guessed by Meshroom), and the known poses (the ones provided by the robot) by id. If list sizes do not match, a warning is provided and the script does not run.

A possible fix would be to perform an inner join of the two lists, then proceeding with the transformation.

In some cases that could be problematic, especially if there are not enough poses, so the optimization would converge to a non-optimal transformation.