Ensure Internal entity exceptions are correctly handled

As discussed here: https://github.com/Azure/azure-functions-durable-python/pull/184

There's an outstanding discussion about what to do when an entity errors out due to an internal exception. For example: what if an operation is missing its name.

One possibility is to propagate the error back to the orchestrator, same as if it were a user-code-level exception. Another approach is to error out completely and let the extension handle it, which is our current approach.

Start by reading the discussion in pull request 184 and the issue's description of missing operation names. Resolve whether internal entity exceptions should propagate to the orchestrator or terminate for extension handling, then define the expected behavior and verify it with focused coverage for that path.

Written by the indexing model from the issue text.