project()'s return shape: the computed output, plus which EvidenceModel field paths fed each output key.
project()
EvidenceModel
Readonly
Per output key, every dotted EvidenceModel path that key's own projector actually read, sorted.
The projection's own output -- exactly what calling the projection directly returns.
project()'s return shape: the computed output, plus whichEvidenceModelfield paths fed each output key.