Support for multiple DataCollectors

[hperrey]

Add support for multiple DataCollector instances running e.g. on different machines:

• reduces network traffic
• reduces local disc I/O

Removal of these bottlenecks is important for running at very high track- and data-rates, e.g. the one-trigger-per-particle operation of the AIDA telescope.

(Potential) issues and ToDo:s:

• Event count verification for all producers
• Online monitoring
• Data file merging for conversion

[hperrey]

Current state of code in the multiple_datacollectors branch: works for command line executables. For example, by starting with

./TestRunControl.exe -l debug
./TestDataCollector.exe -a tcp://55001 -l debug -n testproducer
./TestDataCollector.exe -a tcp://55002 -l debug -n other
./TestProducer.exe -l debug -n testproducer

one can send data from the TestProducer to the DataCollector with the same name. However, in the config, one needs to specify for example

[DataCollector.testproducer]
FilePattern = "../data/run$6R_testproducer$X"

to avoid conflicts when opening the data files.

No tests or adjustments with the GUI have been performed yet.

[hperrey]

The requested feature has been merged with #42; remaining issues with online verification and monitoring are covered by #58.
Will therefore close this issue now.