EN 61000-3-11 and EN 61000-3-12

Presumably it starts with the IEC rather than an EN, the BS-EN is the next adoption and presumably is now a BS.

https://webstore.iec.ch/publication/69084

However it was devised, I cannot see it can be applied in a ground floor bathroom in a granny annexe on a house, the DNO network is not up to it.

So what is the answer, is it don’t fit electric instantaneous water heaters and showers in UK homes?

If it's a heater (so inrush not the issue), then probably it's thryristor controlled, meaning it can have rapid and repetitive changes of load, which gives a worry about supply quality. The  thyristor control of big heating loads tends not to chop the sinusoid shape (which would make strong harmonics, which are more regulated for big loads) but to take a selection of whole cycles or half cycles.  A load that could be tolerable when just on/off could be more disturbing when changing several times in a second. DNOs have long had criteria about impedance for avoiding annoying light changes from simple electric showers (on/off) switching. 
I've always perceived LED lamps as not changing so much as incandescents when the voltage changes, as David says.  They should be able to maintain light output, if designed for that, rather than for dimming. On the other hand I've read of there being some that respond more strongly to small variations. mapj1 mentioned cheap designs in the recent thread about flickering lamps and 220/240V.  Perhaps I've been lucky in getting good quality ones.

If the standard's concern is about effects on other customers in the network, rather than self-inflicted effects in the disturbing-equipment-owner's own installation, then logically the place for the low impedance to be checked is the point of common coupling to other customers, i.e. out on the street main (unless there's a looped service).  There, it's more plausible to fulfil the requirement if it's a biggish net. But you can't easily measure there. Not unless you get help from a neighbour on the right phase, and an extension lead from them onto a voltmeter, and a big load switching several times.

If it's a heater (so inrush not the issue), then probably it's thryristor controlled, meaning it can have rapid and repetitive changes of load, which gives a worry about supply quality. The  thyristor control of big heating loads tends not to chop the sinusoid shape (which would make strong harmonics, which are more regulated for big loads) but to take a selection of whole cycles or half cycles.  A load that could be tolerable when just on/off could be more disturbing when changing several times in a second. DNOs have long had criteria about impedance for avoiding annoying light changes from simple electric showers (on/off) switching. 
I've always perceived LED lamps as not changing so much as incandescents when the voltage changes, as David says.  They should be able to maintain light output, if designed for that, rather than for dimming. On the other hand I've read of there being some that respond more strongly to small variations. mapj1 mentioned cheap designs in the recent thread about flickering lamps and 220/240V.  Perhaps I've been lucky in getting good quality ones.

If the standard's concern is about effects on other customers in the network, rather than self-inflicted effects in the disturbing-equipment-owner's own installation, then logically the place for the low impedance to be checked is the point of common coupling to other customers, i.e. out on the street main (unless there's a looped service).  There, it's more plausible to fulfil the requirement if it's a biggish net. But you can't easily measure there. Not unless you get help from a neighbour on the right phase, and an extension lead from them onto a voltmeter, and a big load switching several times.