Current FifoEvent implementation uses Poco::ActiveResult for async execution.
It has constraint:

• A number of FifoEvents (ActiveResults) depends on maxCapacity in the Poco::ThreadPool and eventually on max-thread-count in the OS if we'll try to increase threads count
• Often FifoEvent (ActiveResult) requires new thread, it is very expensive

My pullrequest contains:

• Separate implementation ActiveThreadPool for ActiveResults & FifoEvents, it uses a fixed number of threads and NotificationQueues in them for Runnable execution. ActiveThreadPool uses round-robin model for task distribution, where task is Runnable
• Some optimizations for notification which replaces copy with move
• Tests for ActiveThreadPool & new test for FifoEvent (testAsyncNotifyBenchmark)
• Changes in project files for including new .cpp & .h and enabling parallel build for cmake projects

This changes allow send as many FifoEvents as you want and execute as many ActiveResults as you want