the problem is in the latest version of numpy1.16.0. They have changed something with flag handling and now this is causing problems in sentinelhub-py.
For now a solution would also be to downgrade numpy to e.g. version 1.15.4. We will make the changes in sentinelhub-py and release a new version soon.
Version 2.5.0 of sentinelhub-py has been released last week. In this version the problem with newest numpy version has been solved. Therefore upgrading sentinelhub-py should be the solution.
By the way, lately new versions of multiple other commonly used Python packages were released. Some of them as a response to the changes in numpy. Therefore upgrading all dependency packages might be a good idea.