Cli runner fails cryptically when the executable for the scheduling system is not in the $PATH

When using the cli plugin runners and the schedulers binary is not in the path, `cli.py` tends fails too cryptically:

```
galaxy.jobs.runners ERROR 2019-10-29 22:43:07,651 [p:20839,w:0,m:2] [ShellRunner.work_thread-0] (2) Unhandled exception calling queue_job
Traceback (most recent call last):
File "lib/galaxy/jobs/runners/__init__.py", line 131, in run_next
method(arg)
File "lib/galaxy/jobs/runners/cli.py", line 105, in queue_job
external_job_id = stdout.strip().split()[-1]
IndexError: list index out of range
```

Probably the cli runners should have a method to check and alert that the required executables are in the path and are executable within the shell where the galaxy process is running.

Start with lib/galaxy/jobs/runners/cli.py, especially queue_job where the empty scheduler output becomes an IndexError. Reproduce the failure with the required executable absent from PATH, then inspect how CLI runners invoke scheduler commands. Done means a missing or non-executable scheduler is reported clearly instead of producing the cryptic traceback.

Written by the indexing model from the issue text.