0.5.1 release

Summary

• Add support for simple ScalarizedObjective Optimization with no outcome constraints.
  Now you can do a simple single objective optimization or a ScalarizedObjective
  optimization by passing different configuration options.
• Metric fetch trial data now passes properties and name to wrapper fetch trial data
• rollback normalize config to previous version
• move metric properties setting to after super
• Change search space parameter name to just parameters again
• Add function to convert wpr prms to boa and func to convert boa prms to wpr
  • function that converts a dictionary with parameters in lots of different section into what boa needs, and prepends them with the json mapping, and then returns the mapping. normalize config can call this, and then saves the mapping. There is also a utility function to undo the mapping as well.
  • Add normalize root mean squared error to metric funcs and add a metric for it