Here is my design. Look at the upper plot pane, where the simulation control signals for the two micro switches are shown. LTSpice uses a switch model that is controlled by a made-up voltage. The switch is closed when the control voltage >0.5V.
Initially, the tube has at least one ball which is holding the bottom switch in the open position, as depicted by the LightBlue trace V(bot). At three seconds, the ball is removed causing the bottom switch to close. That cause the motor to start, dropping a series of balls past the top switch as shown by the Green trace V(top). Eventually, at 12.5s, the tube fills with balls so the last ball holds the top switch closed, shutting off the motor.
I'm using a 555 as a Schmitt trigger with predictable trip points of 4V and 8V. The purpose of the network consisting of R1-4 and C1 is to discriminate between a low state (<4V), an intermediate state, and high state (>8V) of V(rc) Red trace V(rc) in the lower plot pane. When the bottom switch closes (empty stack) this immediately pulls V(rc) low, triggering the 555, causing its output pin 3 to go high, thereby pulling the motor relay. The current through the relay coil is shown by I(L1) purple trace.
To reset the 555 to turn off the motor relay, V(rc) must get above 8V, which happens only when the top micro switch is closed for more than 90% of the time. The pulses caused by the falling balls do not get V(rc) high enough; only when a ball keeps the top switch closed does V(rc) rise high enough to reset the 555. The motor relay should be 12Vdc with a coil resistance of >80Ω.
Note that it takes a while after the top switch comes on steady for the circuit to shut off the motor. You should mount the top switch a little down the tube because an extra ball or two may drop in after the top switch comes on steady.