Not quite a legal requirement in Oz, but virtually so. The standard requires it, deviation from the standard requires high level engineering justification, compliance is simpler than engineering. The problem with fires has been due to shoddy isolators/breakers which didn’t comply with the standards marked on them, and with fake MC4 connectors. They overheat and some of them had non-complaint plastics bodies that did not self extinguish as required. Presumably in RSA unlicensed work is limited to somewhere around 120-145 VDC, so the importance of the isolater/breaker properly carrying the load is even more important due to the higher currents. Should you use double pole, non-polarised thermal breakers for the task, even if the part is deficient, the heat should cause it to open circuit before the smoke gets out. To the best of my knowledge, the notion of removing isolation from the panel array has not been broached to the standards committee (it is a standard, not a regulation, each state has their own law/regulations which reference the standard as a means of compliance). One response incorporated in the standard has been a requirement for sheltering the devices from the weather and UV - IP56NW standards may not have been met by the flaming devices, or, much more likely, it may not be installed properly, compromising the device and letting water in, which doesn’t help. The other thing is the housing deteriorating from UV. Both SA and Oz enjoy low levels of atmospheric pollution, particularly with regard to ozone, so our UV levels are often ten to fifteen higher than Europe, Asia and North America, this damages the plastics IP56 housings - hence the Oz standard requiring non-combustible shelters. In practice, a folded piece of stainless sheet is used.
The following article has a lot of educational pictures at the end showing things going wrong;
http://www.acsolarwarehouse.com/news/solar-fires-dc-arc-faults-on-solar-systems/
You’ll note a lot of MC4 initiated fires. MC4 is Tyco power, not a generic name. The cheap copies of MC4 cause a lot of fires. Both the Aussie standard and the RSA draft require you to use connectors from the same manufacturer on each joint, to reduce the fire risk, many having been caused by the reverse engineered product being not quite the same... If you are worried about fire, worry about this.
You’ll also see a lot of tide lines and rust in failed isolators and switches. Drenching the connections in conductive soup doesn’t end well...
The idea of having isolation where you are working is a very good one, you can keep an eye on it. Decades ago I was in water to my waist down a bore hole installing a 440V pump motor, I had just installed the last wire and was about to put the wiring cover on, when it started turning. The owner had arrived, cut the lock on the isolator with the “Danger do not remove” tag, reinstalled the fuses and turned the pump on. Others haven’t been so lucky, hence the idea of isolation at both ends - it’s a good one. If you can do it with decent fittings that won’t catch fire and protect it from the elements, I’d commend it as a good idea, even if your standards don’t require it. The other reason for isolators on the array is to give firefighters a way to protect themselves, which would be of little use in RSA, as they wouldn’t expect it to be there.
May I suggest the use of doublepole thermal DC CB at your array combiner box? This will help you protect yourself from fools. In my opinion, they are generally are not as good as fuses for dealing with fault currents, so you may prefer fuses as well. But if you can get an IP56 fuse enclosure that protects you from live conductors when you open/close it, that would probably do everything you want. Also, make sure you use drip loops on every cable entry point. Including ones inside or on a protected outside wall - and try to use side or bottom entry to the enclosures. You could also put weep holes at the bottom of conduit drip loops.
Your RSA draft standard looks very well written. Electrical standards are often written in blood, meaning that specific clauses often relate to specific fatal accidents. Thus, although your drafters are clearly very intelligent and must have looked at other jurisdictions to try to learn without exacting a blood price, they may not have anticipated every issue. Don’t confine your thinking exclusively to the standards, lest you contribute blood to a future revision. Your concern about causing fires is very well founded and an inability to isolate at source, should some numbskull break your isolation at the other end likely won’t cause more than inconvenient arcing and equipment damage, but does have the potential to kill you, through another fault being present or through the combination of things you are touching at the time, such as might happen installing a pair of MC4 onto cables.
On some installations I do something a little naughty, where the enclosure isn’t mounted level, I drill a few weep holes on a corner edge that is parallel with the ground. If water gets in, in future decades and no one ever inspects it, instead of accumulating, it has a chance to get out. However, for deniability, not quite getting the sealing 100% correct on a bottom conduit entry can achieve the same thing. I suspect drainage requirements will be a future direction in the standards, the nature of elastomer seals is such that they tend to fail with time. The problem is that so often they only let the water in, not out again.
