Thus, it does not seem to be used as a "maximum speed on the approach", suggering that the AP would not keep trying to reach it, or maybe I'm missing something. Am I understanding the way Target Distance works correctly ?
Well, if you have set a target, then you pretty much always want to finally reach a "target distance" of 0, because you want to get there (If I remember correctly, the only reason to set a value > 0, would be to chase a moving target). By enabling that control element, you tell the autopilot, to brake early enough to stop exactly there, according to which engine is available for braking. This translates into a speed limit as a funtion of distance to target, to not overshoot.
If you are far far away, the AP would be happy to accelerate to any speed until it immediately starts to brake, because of that speed limit. This would be the fastest, but also burns a lot of fuel. As a trade-off you have to decide on a speed at which to coast, and that is the "horizontal speed" setting in that case. As the AP will accelerate to that speed (if possible), and will reduce from that speed in brake phase, I have chosen the phrase "maximum speed", I think.
It is totally possible that the set speed will never be reached, because you already are within a range where that would be too fast and overshoot.
I thing this is not the way the MFD behave. Enabled, Horizontal speed setting is always trying to keep the vessel at this exact value, whether or not Target Distance is activated. In any case, small burst from main or retro are keeping the speed to the set value.
If it is not reducing from the set "horizontal speed", then it seems to be not set up correctly. Because braking to stop at the target should overrule that. Check that you have a target set, that your vessel is able to brake (retro doors) and so on... Sorry for asking "is the power cable plugged in"
Ah, and there is a thing about pointing to the target first. If the AP still tries do achieve that, it is too busy to do any corrections about speed.
The way I used it was : standard deorbit, enable Travel Distance and VS to target, wait until Brake rate = 90%, enable Hor. Speed at current speed, brake.
If you do it like this, none of the settings will have any effect on the AP, until you do the "enable Hor. Speed". This should be indicated by a purple readout, when you enabled them, instead of blue (not quite sure about the VS at the moment). Especially the "wait until Brake rate = 90%" is what the AP should do on its own.
Right after deorbit, I would do (actually started Orbiter after quite a long pause): enable travel distance 0, set horizontal speed to current and enable, enable VS to target. Then just let go.
The only problem with that, is that is messes your speed up a little bit (but will reach target). As you go faster while falling towards Pe, the AP will prevent that from happening and by that reduce your Pe even more. Maybe that is why you waited alle the time without horizontal speed control.
Orbit is actually outside the scope of HoverMFD, so deorbit is quite on the border. It's probably not the best AP for that and taking control and disable AP while watching parameters is probably a good thing.
Maybe David Courtney covers HoverMFD in his recent video series, people seem to ask for. This would be very interesting to me. Not from the point of being most efficient (it probably isn't), but rather to see how people's usage differs from what I had in mind