Perhaps the counterpoint argument is that operating the key switch with method 1 gives a positive indication that the emergency lighting does work for the monthly "flick test", and it also is immediately obvious which luminaires are the emergency ones. I've had several fire officers insist on this over the years.
For a duration test, then it could be argued that this is undertaken at times of low risk so having the main lighting off shouldn't be a problem.
If there are sufficient key test switches (rather than one or two at the DB location, then all you'll be doing is isolating lighting in the room or zone of test.
Method 2 has drawn comments in the past that it's not possible to "see" what is happening with emergency lighting as other luminaires operating at full mains are swamping the relatively low output of emergency conversions.
Ideally, the method of testing should have been considered by the designer (he has that duty under CDM) and agreed with the client and building control - either method being acceptable but with relative advantages and disadvantages in different circumstances.
You'll find all sorts of rules which have no basis in reality such as inclusion of visual indication at the test switch to indicate mains healthy or system in test depending on what day of the week it is, that key switches should have line of site to the luminaires concerned etc etc.
Personally. tend to favour method 1 but with additional key switches to make the are off supply small - but that's just an opinion
Regards
Owen