Automatic testing of remaining jupyter notebooks

Despite #942 being completed, there are still some notebooks that have not been completed due to technical issues. Namely:

### "running-a-refinement.ipynb"
This notebook simply runs too slowly to effectively test right now. Something about this notebook takes up a lot of memory and uses a lot of time to compute. I have tried to test this with an increased cell timeout of 3000 sec, but at least one cell finds a way to exceed this running time. Ideally, we should find a way to streamline or optimise the running of this notebook if possible (maybe the work on #1015 could improve things since this notebook seems to use trajectories a lot). Another solution could be to distribute this time-consuming behaviour amongst multiple cells so that no individual cells time out.

### "molecular-visualization.ipynb" & "read-configurations.ipynb"
These two notebooks utilise the ASE viewer, which requires X11 forwarding (and hence an x-server) to run. While running an x-server itself is not hard on the github runners, an x-server takes up the tty console (apparently due to the x-server needing to control I/O). This prevents the actions runner from progressing, and leaves it "stuck" running the x-server process exclusively. `vnc` has been floated about on forums as a solution to this (https://unix.stackexchange.com/questions/489971/why-is-the-x-server-process-not-a-daemon), since it supposedly runs x-server as a daemon in the background, but I haven't been able to get it working so far. The X3DOM viewer works well on the runners without need for an x-server,

Ideally, there would be some way to "mock" an x-server that runs in the background, just so that MDMC does not throw errors when these notebooks are tested. Alternatively, we can replace ASE with X3DOM in the notebooks for the duration of the tests (perhaps at the nbconvert stage?) or simply leave these out for now.