Update python_virtual_environments.md: insert missing line. When executing this example, there is a conda init error raised if this line was not executed previously.

This reverts merge request !1069

- Currently no such partition available on any cluster

- appropriate styling
- Add description on example
- Adopt from Taurus to Barnard

I see no additional value of this information. It is clearly stated: You
wand to run a GPU application, you need to submit your job to a GPU
cluster.

- Tighten desc.
- Remove "partitions"

- Remove jobname from job file, since it provides no additional value
  here