Adaptive timestep control issue between --continue-on-convergence-error and --min-time-step-size

[locture]

Running models with both --continue-on-convergence-error=true and a given --min-time-step-size (say, --min-time-step-size=1) can cause conflicts and unexpected terminations.

The problem occurs in the following scenario:

1. Simulation started normally.
2. Timestep increased to something larger than initial, say, 3 days.
3. Cannot converge now, so timestep is chopped to something smaller again.
4. Still fails to converge. So the timestep is chopped for another time and it becomes smaller than --min-time-step-size=1 day.

In the scenario above, by setting --continue-on-convergence-error=true I would expect the model to still continue at a timestep of 1 day even though it fails to converge here. However, the simulation actually aborted with this error message:
"Solver failed to converge but timestep "
"{} is smaller or equal to {}\n"
"which is the minimum threshold given "
"by option --solver-min-time-step\n",

It seems that there is a conflict between these logics and the --continue-on-convergence-error=true gets ignored.