The AST dump previously emitted named fields in field-id order, which
made it dependant on registration order and so it could differ between
front-ends. We now emit them in the order declared in the node-types
YAML instead, so that the order is kept stable.
For type checking rules, we need to be able to load schemas (so we know
what to check against). However, since we can't have yeast-macros
depending on yeast (where the schema-handling code currently lives) as
this would introduce a circular dependency, we instead split the
schema-related code into its own yeast-schema crate.
