The attached test case passes, but then makes whatever test run next fail (on the Chromium linux port: we didn't try it on Mac). I walked through this for about an hour and a half with Bernhard, and I'm just confused. The RenderEmbeddedObject renderer is thrown away upon navigation, which makes sense. What doesn't make sense is that it doesn't seem to be the same RenderEmbeddedObject that we mark as unavailable when initially trying to load the plugin. A workaround is to teach the plugin element about it's blocked reason, moving that test from the renderer, which we apparently can't trust, out to the element, which we can.
Created attachment 156928 [details] Passing test which makes the next test fail.
This test does not fail, nor does it cause any other tests to fail after it has run. Perhaps Ander's revamping of the plugin system corrected the problem, or other changes in the loader have resolved the issue.