Fail gracefully if a NekRS run fails in stochastic mode

[aprilnovak]

Reason

When running NekRS via the stochastic tools module, if one of the NekRS runs happens to abort (e.g., from a CFL limit being exceeded), the entire set of stochastic runs terminates. We should see if we can allow a failed NekRS run to fail gracefully (let's see how native MOOSE handles this) without terminating the entire batch of runs.

Design

Perhaps override the converged() method in NekRSProblemBase?

Impact

Facilitate Yiqi's stochastic runs, avoid annoying restarts.