attachIndicator with blocking callback could lead to stale reporting

[swarmshine]

Right now indicator values and all other metrics gathered to a report in same thread.
Given example can lead to blocking whole reporting subsystem:

profiler.attachIndicator("my-blockable-indicator", ()->{
jdbc.execute("query that can block for a long time")
})

Suggestion:

1. Create separate thread pool that gathers indicators values and if it block, it will not affect whole aggregating profiler.
2. Add ability to set value to indicator without callbacks:

profiler.setIndicatorValue("my-indicator", 144)