Your example of the boat doing 50mph has some fatal flaws. Not only would the impeller create a restriction, but the size difference between the intake area and nozzle area would mean that pressures inside the pump would build up to limit volume flow,and remember, 15-20psi is at the front of the impeller, not at the nozzle after the drag of the impeller, stator, tunnel walls etc. Hence only a small volume of water would indeed exit the nozzle while under tow. (It is here that I wish to stress that a small volume is relative to when the motor is running, it would be considered a large volume relative to a garden hose!)
I agree that when you initally mash the gas you will see a tempory reduction of pressure, and when you get off the gas you'll get a spike. However, if you were to couple an HJ212 with 4.0kW turbo impeller to a 496cid HO big block and apply WOT while at rest the pump will simply cavitate. This is due to negative pressure infront of the impeller. Once the boat is making way, the hull and intake design start doing their job there is enough pressure infront of the impeller to ask for WOT and the pump respond accordingly.
To much pressure can be a problem as well. The amount of pressure becomming a problem is dependant on the hull and pump. When pressure tries to escape you'll get what some like to call reversion, this is when the excess pressure blows out of the intake of the pump causing the stern to jump leading to control instability. This is similar to what happens when you get off the gas in a hurry, or the engine seizes. This is why the NJBA (US National Jet Boat Assoc. look after 1/4mile drag boats) recommends the use of BOV's (Blow Off Valves, or pressure relief valves) and break away couplings in an effort to keep the impeller turning and relieve excess pressure out of the top of the pump so the 'pilot' can maintain some resemblance of control.
If the pump were to suck water from below the boat there would be no need for the all important 'spoon' on the bottom of the marathon boats, and the design of the intake would be nearly inconsequential. The many evolutions of design and money poured into this would suggest otherwise.
Waterjets simply don't like to suck, they require a small amount of pressure infront of the impeller to be at their most efficent. Thrust (acting on the boat) from the pump is measured by the pressure difference between inside and outside the thrust face (usually the nozzle) and not the difference between the front and back of the impellers. This is the fundamental difference in thrust behaviour between a propeller and a waterjet.