Add schemas for v2.0

[josee-sabourin]

Set this branch compare to the v1-1 branch to be able to see the diff between the two, I hope that this doesn't get messed up when we merge v1-1.

[barbeau]

@josee-sabourin Looks good! A few comments in-line.

[barbeau]

We should keep all previous valid versions in the enum here, so it should include 1.0, 1.1-RC, 1.1, v2.0-RC in addition to 2.0

[josee-sabourin]

I'm not sure I understand why the versions should be included, this schema would not be representative of a v1.0 dataset, can you explain the reasoning to me?

[barbeau]

I guess it could go either way, depending on how we want to use the schema.

I would thinking primarily in context of generating software to generate, parse or validate feeds from the schema, where you would need to know what the universe of valid version numbers are at a given GBFS version. For example, there should only be breaking schema changes between major versions (e.g., changing booleans from 0/1 to true/false between v1 and v2). So you could use a v1.1 schema to validate, parse and generate a v1.0 feed as well as a v1.1 feed. And if the changes are only semantic (e.g., ID fields must be rotated for floating bikes) and the data types don't change, you could use a single schema (and generated software) to work across multiple major release versions too. So basically it would potentially be less software to maintain for people generating software from the schema.

But you could also make the argument that we want to keep a strict 1:1 mapping between the schema and it's version, so it can only be used for the same version feeds.

Maybe a question for the community and how they plan to use the schemas? In the meantime it's fine if you want to keep it showing only the latest version.