TitelExcessive Agent output can interfere with check_mk

Currently check_mk will always try to process all agent output. If the agent produces a lot of output, some checks may cause excessive cpu load and memory usage, effectively making the monitoring system unusable. To the user this will appear like a bug since the amount of data between agent and monitoring system is not visualised and not warned about so there is no way for the user to find the cause of the slowness. To remedy this we could add readings and warn/crit levels on the agent output size, agent response time, execution time and memory usage. In addition we should add support for memory profiling of individual checks so that problematic checks are easier to identify.