This is done for both blocking and non-blocking executor in ThreadUtils
to avoid default creation of 2*CORES threads to mainly idle in the
background which becomes even more non-sensible when there are 24, 32,
64, etc. cores.

Also support system property specification of the amount of non-blocking
thread pool core size.

The old legacy system property simantics.executor.blockingMaxThreads is
also still supported but only used if the new replacement property
simantics.executor.blocking.corePoolSize is not used.

gitlab #678