Thank you for providing the update. I tested this build on Windows using Gaussian 16 .out files.
While the previous file loading error no longer occurred, covalent bonds were not rendered, as shown in the attached screenshot. On the other hand, .fchk files continue to load properly as before.
It looks like the water was parsed as if the coordinates were in Bohr instead of Angstrom. That would generally indicate that it picked a different file format again (i.e., not Gaussian)
Can you please try renaming the file to something like exam.g16 to see if it’s a problem with the format guessing code?