Refactor run workflow

[alecandido]

The run workflow it's whatever is currently inside db_interface.py, since that file it's managing the full benchmarks runtime.

• split the current db_interface.py in:
  • a benchmark-runner library (into banana)
  • program-specific tasks (into yadmark), like collecting the comparison (and maybe immediately dumping into a binary blob)
• split regression module