Actually respect NUPCOL (not NUPCOL+1) and allow network to converge with GLO
Also:
- Some more informative output in case network does not converge
- Change order of return arguments in updateWellControlsAndNetworkIteration to be consistent with updateWellControls
jenkins build this failure_report please
Note. I have a PR where this code is refactored. https://github.com/OPM/opm-simulators/pull/6095 The refactoring also contains more info if the network does not converge.
jenkins build this failure_report please
Test failures are mainly due to different time-stepping. I've been through all the failed tests with shorter time-steps and find results to be (visually) identical to master, except for two tests that differ in a single time-step - these become identical to master when using NUPCOL=11.
Thanks for looking into the test failures. I will start the process of updating the reference results and merging this.
jenkins build this update_data please
jenkins build this opm-tests=1322 please