Skipping to next optimizer's iteration if MDA is not convergent after the max of iterations

Hello,

I am using Gemseo v5.3.1 and I've a question about the behaviour of optimization algorithm and MDA solvers in a MDF formulation. In my study case I have two MDAs in the MDO process, the optimizer starts the first iteration, in the log I can see that differents MDAs are ran but the residual is too high and it finally ends up in NaN values so the algorithm stops. I expect that the algorithm runs a second iteration after that the first two MDAs loops of the solver, but it is not the case. The solvers seems to be stuck in the MDAs that are not convergent and the optimizer cannot keep on exploring the design space.

Here it is an example of my log (I use a MDAChain with 2 MDAs GaussSeidel):

 WARNING - 10:29:16: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 4.873030204723072 is still above the tolerance 1e-06.
 WARNING - 10:29:23: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 2.0468813237207883 is still above the tolerance 1e-06.
 WARNING - 10:29:30: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 4.811908384915084 is still above the tolerance 1e-06.
 WARNING - 10:29:37: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 4.775499439627299 is still above the tolerance 1e-06.
 WARNING - 10:29:44: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 0.16837510345704895 is still above the tolerance 1e-06.
 WARNING - 10:29:50: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 8.686721319598869 is still above the tolerance 1e-06.
 WARNING - 10:29:57: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 1.5505959497248465 is still above the tolerance 1e-06.
 WARNING - 10:30:04: MDAGaussSeidel has reached its maximum number of iterations but the normed residual 4.775502595461469 is still above the tolerance 1e-06.
...

Is there a reason behind that ? Is it a beahviour that can be deactivated in order to keep on with the optimization ?