Introduce flags on #Lineage that result in additional, standardized constraints on lineage validity

[sdboyer]

#Lineage.joinSchema provides one mechanism for constraining what's acceptable as a schema in a lineage.

It might be worth introducing another, parallel system that allows the expression of additional constraints. The lineage author-facing interface to these might be a series of named boolean flags at the root level. Something like this:

#Lineage: {
    ...
    constraints: {
        // All schema in this lineage must rely on the subset of CUE that is precisely translatable to OpenAPI.
        openapi: bool | *false
    }
}