jwnimmer-tri
commented
2 days ago From f2f -- this approach was overkill, because as it turns out calculating reaction forces is cheap, so the better answer is to do it on every time step, even if nobody is pulling on the output port. So, we will discard this PR.
A few parts here were small clean-ups for clarity, which have been split off into #21660 and #21661 to follow-up with later on. (Those PR's are fully cleaned up yet.)
Towards #21623 and therefore also #20545.
This change is