Improve the handling of config syntax errors during service start.

From a Customer SFDC case

> We had a syntax error in logstash.yml. However, when started as a service, there's no indication of this, as the service falls back to default settings and ignores the logstash.yml if there's a config error
As such, it was running with no explicit config, which stopped it logging. I only discovered this by running logstash manually on the command line and checking the output
It's sorted now and stable, but I do think this is a very dangerous way of operating. I've never dealt with a linux service before that would start up with a config error, they would all report an issue and fail to start. Continuing to start with an unexpected config is poor.
FYI we're running using the 'yum' install and starting with 'initctl start logstash'

With the coming changes to the pipeline execution can we bear in mind running LS as a service when completing said features.

- Version: 5.X
- OS: Amazon Linux, RPM install