Performance issues with the charm

[simskij]

@javacruft seems like performance has been decreasing lately, we can consider merging this work, what's the status?

Originally posted by @PietroPasotti in #197 (comment)

Seems like we still have performance issues where Traefik takes a lot of time to settle. @javacruft should be able to give more details on what they are seeing.

[PietroPasotti]

I have been told by the sunbeam folks that @javacruft 's recent PR caching lightkube calls already resulted in a major improvement, making traefik approximately as fast as all other cos-lite charms to settle.

That said, as part of #269 I am taking a shot at some workload pebble api call optimizations that should also make traefik quicker. Let's see how far that gets us.