Convergence stagnation at 1e-4 in GCMR flow blockage transient

[GiudGiud]

Bug Description

In the new setup (post fixes and migration to general field transfers), the transient flow blockage GCMR inputs do not seem to converge past 1e-4 on the richardson iteration (multiapp iteration) for the first time step. Maybe with more iterations it might go lower but i have not checked
This is being discovered as the inputs are finally tested using HPC

Steps to Reproduce

Run the GCMR inputs being modified in this PR: #506

Impact

Likely a cause for concern in the correctness of the result with the updated codes and inputs.
Warrants further investigation.
This limited convergence is still better than the current state of the inputs (non functional due to minor issues / untestable)

[GiudGiud]

@miaoyinb @aabdelhameedd for awareness

[miaoyinb]

Thanks, we will take a look.