More generalizable summary interfaces

[oesteban]

It is becoming more apparent that a more generalizable code for summary interfaces is necessary.

Additionally, in the context of multi-site studies, we should have more flexibility to add more metadata details such as the scanner vendor, model and software version.

This issue, therefore, implies:

• [ ] Centralizing summary interfaces in niworkflows (i.e., remove the code from fMRIPrep and dMRIPrep, so that just setting a template would suffice).
• [ ] Flexibilizing the inputs of the interface - i.e. parse the template to make out the inputs of a DynamicInputTrait specification.
• [ ] Implement a language/protocol to tell these interfaces when to pull information out of the BIDS metadata.