Dear Developers,

I have encountered a strange issue when writing trajectories from a non-orthogonal box in the netcdf format.
I find spurious correlations in the rdf functions I calculate from my trajectories, which make me believe that the cell info is not read correctly. I could reproduce the issue using both OVITO and MDanalysis.
Please find attached the trajectory in the .extyz format written by GPUMD which I believe is correct and the .netcdf format. I had to rename the file extensions into .txt to be accepted from github, (dump.txt is the extxyz and movie_4.txt the netcdf)
dump.txt
movie_4.txt