C++ tests are being included in the report themselves, which skews overall results with misleading coverage. We want the coverage of the source files that the tests exercised, not necessarily the lines within the test file.
These should be filtered out.