The limit depends on how much current the alarm passes through the contacts (and the resistance of their wiring) and at what total voltage drop across that circuit, it decides that some switch is open. For instance, if the switch loop is fed with 10 mA and decides that an open condition is indicated at a voltage drop of 2 volts, then the total resistance of the loop and all the switch contacts must drop less than 2 volts while 10 mA passes through them. The limiting resistance would be 2/.01= 200 ohms. For any reasonable gauge wire, that is a lot of length, but this example is completely hypothetical.
I would want to find out the limiting resistance by test, using a variable resistor in place of the loop, and measuring the resistance that is interpreted as an open, by experiment. Then, I would try not to make any loop that had more than a minor fraction (say, 1/3rd) of that resistance, to allow a little room for the contact resistances to rise with oxidation. It is also a good idea to wire the loop as a flattened one, using a pair of wires that take the same route in both directions, to keep the loop from acting like a magnetic loop antenna that will cause false alarms any time a thunder storm passes nearby.