[Bug]: propagation of usage flags makes base model type usage ambiguous

[jhendrixMSFT]

Describe the bug

Some change in 0.47.x causes usage flags to propagate from derived type to base type. This means that it's no longer possible to identify base types that aren't explicitly annotated.

For Go, which does not have inheritance, we omit base types by default unless they're explicitly referenced (directly or transitively) via an operation.

For model-only builds we can only rely on usage flags to control the output. Now that these flags are propagated, the usage on base model types is ambiguous.

Reproduction

For the following tsp, BaseClassThatsPruned has a usage of output even though it's not explicitly annotated as such.

import "@azure-tools/typespec-client-generator-core";

using Azure.ClientGenerator.Core;

@service({
  title: "ModelsOnly",
})
namespace ModelsOnly;

model BaseClassThatsPruned {
    id: int32;
}

model DerivedOne extends BaseClassThatsPruned {
    name: string;
}

model PreservedBaseClass {
    id: int32;
}

model DerivedTwo extends PreservedBaseClass {
    name: string;
}

@@access(ModelsOnly.PreservedBaseClass, Access.public);
@@access(ModelsOnly.DerivedOne, Access.public);
@@access(ModelsOnly.DerivedTwo, Access.public);
@@usage(ModelsOnly.PreservedBaseClass, Usage.output);
@@usage(ModelsOnly.DerivedOne, Usage.output);
@@usage(ModelsOnly.DerivedTwo, Usage.output);

Checklist

• Follow our Code of Conduct
• Check that this issue is about the Azure libraries for typespec. For bug in the typespec language or core libraries file it in the TypeSpec repo
• Check that there isn't already an issue that request the same bug to avoid creating a duplicate.
• The provided reproduction is a minimal reproducible example of the bug.