Limit overhead of trace_methods

When applying the [`trace_methods`](https://www.elastic.co/guide/en/apm/agent/java/current/config-core.html#config-trace-methods) configuration option to a large number of methods or event to just one method that is called very often, the performance of the application can severely suffer.

This is a proposal to de-instrument methods that are executing very fast. A good default could be that if a method ever takes less than 10µs, it will be queued for being de-instrumented.

The agent regularly checks the queue and if it's not empty, it un-instruments the registered methods.

There is still overhead involved in instrumenting and de-instrumenting methods and it doesn't solve the underlying problem with trace_methods. Our recommendation would still be to use profiler-inferred spans monitoring a whole codebase in production. But it could make `trace_methods` less dangerous to use.