It's difficult to come up with a safe way of working that eliminates shock risk if you remove a neutral from the neutral bar (due to possible borrowed neutral), without making other circuits dead to test for it (by insulation resistance).
For example: replace light fitting in upstairs bedroom. Isolate upstairs light breaker, lock off, prove dead (at the breaker you've just isolated, and also for good measure at the light fitting itself before you start taking wires out). Leave downstairs lights on - you note the landing light is fed from the downstairs circuit, provides a bit of ambient light from outside the bedroom.
Unknown to you, althrough the landing light is fed from downstairs lighting circuit (live), return neutral is via upstairs lights. Neutral is firmly attached to earth (on a TN system at least) at the board, but what if the bedroom light fitting you're replacing is between the landing light and the DB? As you separate the neutrals, you might frown, wondering if it was a coincidence that the landing light went off at that point, maybe you didn't even notice. At this point, the neutral tied to the neutral bar will be at zero volts (the same potential as earth), whilst the one attached to the landing light will be at 230V.
If you touch the (now live) neutral, and also the cpc, then your chances of getting a bad shock will depend on (a) whether there's an RCD fitted (both lighting circuits would need to be on the same RCD for them to work at all), but also (in the absence of an RCD), interestingly, what sort of lamp is in the landing light. If it's an LED, you might not even notice it, or just experience a little tingle. These pull very little current, a few mA - which means they have a high resistance, higher than your body, so most of the voltage will drop across the LED. If it's a CFL, then you would experience a noticable tingle, a reasonable shock - depending on your body's resistance own resistance, it might be the equivalent of 100-180V. The worst case is if it's a high wattage tungsten/halogen. There will be very little voltage dropped across this, so you will get a substantial belt.
Or perhaps you never even get as far as getting to the light. You isolate the upstairs circuit, lock off, prove dead. You want to ensure the circuit is suitable for alterations. You've already done a quick Zs at the light prior to locking off, to show there's an earth (and disconnection times are ok). Since you're at the board locking off, why not do a quick insulation resistance test? Of course that will mean disconnecting the neutral for the upstairs lighting circuit from the neutral bar. The moment you do so, it's a live conductor, and a shock risk.
How to eliminate this risk? Well, unless you (more likely, the customer) is happy for you to turn everything off while you do the work, it's tricky. You could make everything dead, remove the neutral of the circuit you want to work on from the neutral bar, then test the insulation resistance between this (disconnected) neutral, and the rest of the circuits still connected to the neutral bar. But not only does the "suspect" load (the landing light) need to be on, there also needs to be other loads on too - have a look at KEV1N's drawing in post #8: Remember, everything is dead at this point, we have isolated the whole board at the main switch, we want to work on circuit 2 (top) so we have removed that N from the neutral bar. To get a low IR reading that would tell us there is a problem, not only would the landing light need to be turned on, one or more of the lights in circuit 1 would also need to be connected. I suppose you could measure IR between the N you have disconnected, and L of every other circuit apart from the one you're working on... but the landing light (if that's the culprit) would still need to be on for it to show up.
So yes. Tricky.
