it is indeed unfortunate that it took several years before there was serious consideration to the impact of site-local on networks, DNS, and applications. I view this as an indication of serious failures in our process - IPv6 was approved (under pressure, it must be said) by the working group and IESG despite serious technical omissions. fortunately proposed standards are not carved in stone, and these particular omissions are easy to fix.
There is currently no RFC that defines the usage model for IPv6 site-local addressing. The prefix is defined in the addressing architecture RFC (currently at PS), but the usage model is defined in the scoped addressing architecture (a WG I-D).
This I-D has been relatively stable for a couple of years, but has not gone to the IESG, mostly due to IPv6 WG deadlock over the architectural issues regarding site-local addressing. During that time, we have been slowly working through the various architectural issues caused by site-local addressing, and building WG consensus regarding what to do about it.
So, I think that this is actually a case of the WG process working properly (if a bit too slowly).
Margaret