I’m not sure the original question was answered. Here’s my attempt.
FTP is an estimate of the power you can produce at your lactate threshold (more precisely at LT2, aka maximal lactate steady state). Go below your FTP and you can sustain that for some time. Go above it and you cannot sustain that effort. This is the one we estimate from a ramp test, a 20 min test protocol (which has a 5 min maximal effort, then 5 mins rest, then the infamous 20 min effort, and if you do this as per protocol it truly sucks). Or as a shortcut, Zwift will estimate it from your 20 min power. That might be a fresh 20 min effort, that might have been from a 19 mins 30s effort after which you soft pedaled, it might have been from a 21 min effort, it might have been from a race where you were staying with the pack but the pack was very much not going at a constant effort. It’s a bit more of a dirty estimate.
zFTP and zMAP are different. Critical power (CP) models are mathematical models that try to predict your maximum power from about 2 to about 30-40 minute durations. As input, they need two fresh maximal efforts, one about 3-8 mins and the other about 12-20 minutes. Consider the graph here. The CP model assumes a hyperbolic relationship between the two input efforts. The asymptote is known as the critical power - it’s the dashed line in the graph, although that overlaps with the solid line of my then-estimated FTP.
zFTP is the critical power estimated by the model, and it’s interpreted similarly to traditional FTP: it’s an estimate of the max power you can sustain for a long duration. It’s usually close to traditional FTP, but they aren’t mathematically equivalent, I believe. I think that zMAP may be the model’s estimate of your 5 min fresh max power.
If you have not done a real maximum short-term effort, then it will assume your max potential power is a hyperbola with a flat slope. This might actually underestimate your zFTP. Obviously a new maximum 20, 30, or 40 min effort would raise the model’s estimate of your CP.