As of two days ago, the nutils-of test has started generating slightly different results, causing the comparison (and subsequently the test) to fail. This PR updates the reference data to comply with these new results, as they appear to be consistent between runs (see ŧhis example run with adjusted data).
Note that this does not completely fix the issue, more so we should try to track down what causes this test in particular to act so differently from others. We previously observed this test to generate different results when run on TravisCI compared to a local machine, even though both instances are done in identical docker containers derived from the same images. The recent change of result data might be related to this previous occurrence.
As of two days ago, the nutils-of test has started generating slightly different results, causing the comparison (and subsequently the test) to fail. This PR updates the reference data to comply with these new results, as they appear to be consistent between runs (see ŧhis example run with adjusted data).
Note that this does not completely fix the issue, more so we should try to track down what causes this test in particular to act so differently from others. We previously observed this test to generate different results when run on TravisCI compared to a local machine, even though both instances are done in identical docker containers derived from the same images. The recent change of result data might be related to this previous occurrence.