The v3 scenarios_t.py tests need to be refactored.

Currently, many of the tests are testing multiple things in a single call. Not only does that make interpreting a fail more difficult, but it means that if one of the things (durability in this case) isn't supported on all platforms, all the other things are also not tested on that platform since we skip the test.

So - test_multi and test_mutatein are examples, but probably the rest of the tests are similarly structured. This was converted from v2 tests, and so this will happen with the others that we convert as well. But this is already in the V3 (so part of the integration tests), lets fix it sooner rather than later. Put in 3.0.0 for now.