If one needs greater control over how the different worker processes are
started and managed at runtime the "celery multi" command cannot
reliably be used.
So now there are three different sub commands for celerycmdline:
* nodes: prints the ids of the different celery workers that need to be
started to fulfill all queue constraints
* start: prints the "celery worker" commands to start the workers
* rate_limit: prints the celery commands to apply rate limit settings to
For every command the "-n" parameter can specify a number of node ids
for which the appropriate commands should be printed.