Test: cts: Drop cts script

The remaining functionality is not particularly useful.

• It creates a DSH (distributed shell) group of cluster nodes, so that the group name can then be passed to cts-lab and helper scripts.
  • The active developers don't use DSH and don't use the helper scripts directly. It's not difficult to pass an explicit list of nodes to cts-lab.
  • If there is ever demand to bring this type of feature back, we could consider a CTS configuration file that stores (among other things) a cluster-name-to-node-list mapping. That could be convenient for using a single exerciser with multiple test clusters. However, that's likely to be a rare use case, and any convenience likely doesn't justify the bloat and maintenance.
• The [--]clean option:
  • Removes the CTS log file. The file name is a hard-coded template that changes only based on the cluster name. In general, it should not exist unless this very script created it.
  • Stops/kills cluster services on all nodes. Other high-level tools like pcs and crmsh can do this.
  • Kills valgrind on all nodes (even if it's running something unrelated to Pacemaker).
  • Deletes cluster logs and a bunch of system logs that are unrelated to Pacemaker, on all nodes.
  • Deletes Pacemaker state and shared memory files on all nodes.
  • Deletes all "*.valgrind" files in /tmp on all nodes.
  • Restarts rsyslog and the systemd journal on all nodes.
  • Overall, this option seems overzealous, affecting things that are not owned by Pacemaker.
• The [--]boot option performs the cleaning described above and then simply starts the cluster on all nodes. Other high-level tools like pcs and crmsh can do this.
• The [--]run option performs the cleaning described above and then runs cts-lab, for 500 iterations by default, with any extra arguments that were given. That doesn't provide significant added value or convenience vs. running cts-lab directly.

Signed-off-by: Reid Wahl <nrwahl@protonmail.com>