create SST benchmarks for scaling of trivial configurations

[bhpayne]

Why this matters:

• regression comparison when new SST versions are released
• evaluation of impact of SST changes on certain aspects of models in use
• identifying scalability bottlenecks (e.g., memory, compute, MPI latency
• determining the impact of feature changes to sst-core, e.g., the cost of profiling

For example, https://github.com/researcherben/structural-simulation-toolkit/tree/master/sst_scaling/ring

[researcherben]

To measure:

• scaling of ring: https://github.com/researcherben/structural-simulation-toolkit/tree/master/sst_scaling/ring
• impact of global parameters: https://github.com/sstsimulator/sst-core/issues/756
• serial scaling of each model
• parallel (MPI) scaling of each model