Decide on convergence when using Picard iterations

[moosebuild]

Imported from https://github.inl.gov/ncrc/subchannel/issues/58 : @moose-ncrc created issue at Feb 21, 2022 09:12AM MST:

In GitLab by @andrsd on Feb 21, 2022, 09:12

Description of the enhancement

Compute residual for p, T. Use that to decide on convergence of a Picard (fixed point) iteration.

Rationale for the enhancement

When doing Picard iterations, MOOSE considers only primary variables, thus it reports 0. for subchannel solves. If we can feed the residual from subchannel into the Executioner, then we can control convergence criteria as in a usual MOOSE-based app.

Identified impact

(i.e. Internal object changes, limited interface changes, public API change, or a list of specific applications impacted)

Gain functionality, better usability.

Notes

• Since subchannel use p and T, this may be a little more challenging to implement, since MOOSE decides on convergence using only a single value.