[alpaka] Initialise only the selected backends

[fwyzard]

Only the backends enabled at build time are supported in the command line options, and listed in the help message. Only the backends selected on the command line are initialised.

[fwyzard]

Building alpaka without CUDA support now results in

$ ./alpaka --cuda
Invalid parameter --cuda

./alpaka: [--serial] [--tbb] [--numberOfThreads NT] [--numberOfStreams NS] [--maxEvents ME] [--data PATH] [--transfer] [--validation]

Options
 --serial            Use CPU Serial backend
 --tbb               Use CPU TBB backend
 --numberOfThreads   Number of threads to use (default 1, use 0 to use all CPU cores)
 --numberOfStreams   Number of concurrent events (default 0 = numberOfThreads)
 --maxEvents         Number of events to process (default -1 for all events in the input file)
 --runForMinutes     Continue processing the set of 1000 events until this many minutes have passed (default -1 for disabled; conflicts with --maxEvents)
 --data              Path to the 'data' directory (default 'data' in the directory of the executable)
 --transfer          Transfer results from GPU to CPU (default is to leave them on GPU)
 --validation        Run (rudimentary) validation at the end (implies --transfer)
 --histogram         Produce histograms at the end (implies --transfer)
 --empty             Ignore all producers (for testing only)

[fwyzard]

Building alpaka with CUDA enabled, but running with e.g. the serial backend, CUDA is never initialised:

$ compute-sanitizer --require-cuda-init yes --tool memcheck ./serial
========= COMPUTE-SANITIZER
Processing 1000 events, of which 1 concurrently, with 1 threads.
========= Error: No attachable process found. compute-sanitizer timed-out.
========= Default timeout can be adjusted with --launch-timeout. Awaiting target completion.
Processed 1000 events in 3.383860e+01 seconds, throughput 29.552 events/s.