I have zero control over Hockmir, per your please get opinion also. . .
and free account so no PM
That would be one very large capacitor to have much impact. Also the thing with a capacitor, like starting capacitor on a motor, is to give a quick discharge, short-term boost. So while you could use capacitor to try to get ability to use a smaller amperage line, once the capacitor had discharged its draw would be additive to that of the rest of the unit and continue when the unit was off. The gain would be in reduced electrician install costs, try to get to that magic "plug-in" point of use unit we all seem to want to be able to buy and to see sold. Since the magnetrons draw at a particular level while on and they would be on while the water is flowing it seems that there would be little in the way of peaks that the capacitor could help to level out.
Here in AZ people are big on swamp-coolers, which are basically big squirrel cage fans that blow air through wet pads to chill the dry air, so there is minor control switching and just one half horse to one horse motor in residential scale models. A friend of mine showed me how he had hooked up a capacitor in the circuits and claimed that the result was that the power consumption of the swamp cooler went way way down. I do not doubt him, but don't fully understand why, or for that matter why then swamp-coolers are not sold that way (since they are basically a money saving product sold to allow people to keep the air conditioner shut off for the majority of the year). The way it was explained to me was that the capacitor allowed the motor to get past continuous stream of little peaks in draw as it moves the imperfectly balanced cage fan around, cutting out many little moments of inefficient line current draw.
