At least the failing tests that fail with ... #     doesn't match '(?-xism:^(404|500)$)' are because my test case is too specific. The next release will have relaxed test cases that check for /5\d\d/ instead of 500.

The other failure happens because the test loads iframes that are not found, and your proxy substitutes the failure and subsequent non-existent documents with some made-up documents. I guess I will have to make the tests more resilient here as well, by substituting the non-existing frames with (empty) documents that can be found.