Hasura is designed to implement the entirety of the Business Logic Layer - it provides an interface directly on top of Postgres. Certainly this isn't appropriate for certain applications, though https://hasura.io/blog/custom-business-logic/ makes it very versatile. But for many line-of-business applications it's ideal.
That only works if your API access is all or nothing. The existing authorization tooling will just see everything as requests to `/graphql`. If you want per-query or even per-field permissions you need to do something inside the GraphQL layer.
Why does Hasura need to know about the authorization model? Graphql.org's best practices guide[1] says authorization should be in the business layer and should be a single source of truth. In other words, GraphQL responses should be going through the same authorization module as REST or RPC responses.
[1] https://graphql.org/learn/thinking-in-graphs/#business-logic...