Oscillating reference results in case 0B_WELPI_MODEL2

[GitPaean]

When testing and reviewing the PR https://github.com/OPM/opm-simulators/pull/4590

Checking the regression failures was the major effort comparing with the code review.

One of the failed cases is the case 0B_WELPI_MODEL2 . When running with master branch and TUNING or small time steps, we will see the oscillation result for one of the wells, The well is under RESV control if not mistaken.

When we have this type of reference behavoir, it means any change in the time stepping will make the regression test fail.

The code in the PR https://github.com/OPM/opm-simulators/pull/4590 looks fine, and investigation of the setup of this testing case will require some substantial efforts, which we do not have, . Posting the issue here, hopefully we can revisit the issue later. And also this test failure should not be in the way of the PR https://github.com/OPM/opm-simulators/pull/4590