Improve externalization of internal data

To be able to define assertions on internal data, we make the relevant `OutputWires` available and hook into them. This is, of course, not ideal. Let's try to come up with a better solution.

Start by tracing the OutputWires mentioned in the issue and how they are exposed for assertions on internal data. The issue does not name files or tests, so first locate the current externalization and assertion hooks. Done means providing a better solution that avoids making the relevant OutputWires directly available while preserving those assertions.

Written by the indexing model from the issue text.