The problem seems to be that connecting a length of open wire to the sensor terminals is sufficient to create a feedback loop and cause the timer to pulsate the watering. Without knowing the circuitry behind the sensor terminals it is difficult to propose the best solution. Using a wireless sensor, as I am, the work around is to add a 0.1 uf cap across the timer sensor terminals. This seems to isolate the 18″ of wires to the rain sensor base. If someone is using a wired sensor with many feet of wire between the timer and the rain gauge, a capacitor may not work. As mentioned, this s a hardware issue and may not be able to be solved with firmware update.
RCC344, are you using a shielded cable between the the timer and the sensor receiver and is the shielding attached to ground? It might help reduce noise that is being picked up by the wires. for the benefit if others, what RCC344 has done is add a pass-thru filter to to the input of the sensor terminals. His allows noise spikes on the sense wires to effectively pass through the capacitor instead of going through the timer (and changing the state of the timer).
I am using a Hunter wireless sensor and do not have the issues you encountered.