[GeoMechanicsApplication] Bug in time step mechanism C++ workflow

[mcgicjn2]

Given the attached project in the following comment, I expected a single linear calculation to take place it is a K0 step.

The output confirms that it has converged at the end time step, i.e. -0.2

However the stage fails with the following error:

KratosGeoSettlement: Error: Delta time (2.77556e-17) is smaller than default minimum allowable value 1e-05

It should also be noted that the input start time contains a rounding error: thus startTime = -0.30000000000000004.

[mcgicjn2]

RecreateIssue.zip

[rfaasse]

The issue can be replicated in a unit test by providing the following state.

KRATOS_TEST_CASE_IN_SUITE(ScaleIncrementToAvoidExtraSmallTimeStep, KratosGeoMechanicsFastSuiteWithoutKernel)
{
    AdaptiveTimeIncrementorSettings settings; // with EndTime = 8.0
    settings.StartTime = -0.30000000000000004;
    settings.EndTime= -0.2;
    settings.StartIncrement          = 0.1;
    auto time_incrementor            = MakeAdaptiveTimeIncrementor(settings);
    auto previous_state              = TimeStepEndState{};
    previous_state.convergence_state = TimeStepEndState::ConvergenceState::converged;
    previous_state.time = -0.20000000000000004;

    time_incrementor.PostTimeStepExecution(previous_state);
    KRATOS_EXPECT_DOUBLE_EQ(-0.2, time_incrementor.GetIncrement());
}

We need to check if this state should not occur (the previous_state.time could be seen as incorrect), or if the PostTimeStepExecution function should be able to handle it correctly, and not throw the reported error.

[rfaasse]

FYI, A similar error is not generated for the python workflow (so there it's correct)