Added protocol buffer models

[margaritiko]

Add protocol buffer models for representing bindings, dependencies and graph

All models constructed with proto2. This is preferred because in proto3 all fields are optional which can add ambiguity.

• Binding is represented by its unique string-key.

• Dependency model represents an edge in graph which always contains target node.

• BindingGraph contains a map adjacency_list which simply matches binding or component (access it by string representation of key) to list with other bindings.

[margaritiko]

All these files with Proto suffix are generated automatically, so don't waste your time on them ⏰

[kwchan1995]

How about excluding generated code from source control, by marking them in .gitignore? And maybe add a command for the code generation in the README. This way it will reduce clutter and prevent the generated / actual source code from ever going out of sync :)

[margaritiko]

Cool, thanks! I think it's good idea, I will add entire autogen folder into .gitignore

[kwchan1995]

Perhaps controversial, actually the internal guidance is to prefer optional protobuf fields to required, because it makes future refactorings easier without having to support the required fields indefinitely (see https://developers.google.com/protocol-buffers/docs/proto#specifying_field_rules and protocolbuffers/protobuf#2497 (comment)). This is also why required was removed in proto3. Would it be possible to move to proto3 for future-proofing if not too much work please?

[Ratchette]

Adding this documentation to the readme is a really good idea.