Separate call for running a fully-specified pipeline?

In the current API, the way to run a fully-specified pipeline is to give it as 'template' to SearchSolutions: core.proto#L147-152

Apparently TA3s don't like this too much, and looking at it, it's a bit awkward because:

• You still have to go through all the rest of the calls: SearchSolutions, GetSearchSolutionsResults, FitSolution, GetFitSolutionResults, ProduceSolution. That's a lot of callbacks/streams.
• You have to provide a problem when calling search, which might not be relevant if you're preparing data before the user has formulated a problem.
• The solution_id is useless for anything but FitSolution anyway. Calling ScoreSolution on it would fail if it's not a prediction pipeline. SolutionExport probably as well.
• Because of that, it is not really a "solution". It is awkward for the TA2 to have to hold this in its "solution ID" namespace when it has no score associated with it, etc.

It might make sense to have a single call that goes from fully-specified-pipeline to ProduceSolutionResponse, or at least to fitted_solution_id.