Allow the Elasticsearch exporter to be reconfigured after the engine has started

[RomanJRW]

In Optimize, we segregate Zeebe data per test by applying an index prefix to the exported data unique to a given test. With EZE, the configuration must be supplied up front, meaning that an engine instance can't be shared across classes with a dynamic exporter prefix, like we currently do. It also means that we might not be able to reliably/easily clean up the data specific to a given test, as it can't be identified by its prefix.

If EZE is performant enough, then we are also completely open to having a new EZE instance started per test rather than per class, with the prefix determined up front at the start of the test.

[Zelldon]

Thanks for opening this.

If EZE is performant enough, then we are also completely open to having a new EZE instance started per test rather than per class, with the prefix determined up front at the start of the test.

Especially to this I can say it should be fast enough to use always a new one. See https://github.com/Zelldon/eze-sample

[RomanJRW]

@Zelldon - benchmarks look good, thanks for the link 👍. We're coming from a place of quite bad performance at the mo with how our tests are set up, so this would potentially be a very nice improvement. I think the prefixing is the bigger issue here then, to allow us to isolate the data for a given test