Using +ppn instead of ++ppn should not be silently ignored by netlrts layer

[jcphill]

If a netlrts-smp layer is launched using ++n with +ppn instead of ++ppn, e.g., charmrun ++n 6 ++mpiexec ++remote-shell srun namd2 +ppn 8 then the +ppn 8 is silently ignored and the user gets the equivalent of +ppn 1. NAMD will complain about smp binaries running with ppn=1 except in the case where the number of replicas is equal to the number of processes (which is common for GPU runs and actually required for NAMD 3). An unexpected +ppn argument that doesn't match the actual ppn should be an error.

Related to https://github.com/UIUC-PPL/charm/issues/2897