[OPS] Add diagnostics script for shell nodes
We should have a way to test if our shell nodes are fully functional. A diagnostics script that can be run regularly.
The script should run some basic diagnostics, like:
a) try to arv keep get something
b) try to arv keep put something
c) try to launch a diagnostics job on the cluster (?)
When all is well, the script should produce no output and exit with 0. If something goes wrong, it should not suppress any stdout or stderr and exit with non-zero.
The script will be run via cron.