If you are talking about variants inside of syntax, you are right. There is only need for ekavian/iyekavian manual transliteration, so there is no need for ((sr-ec::...\sr-el::...\ etc.)). We need only two variants: left should be ekavian, right should be iyekavian. So, in general, we need only ((<ekavian>\<iyekavian)). Left variant should be sr-ec and sr-el; right variant should be sr-jc and sr-jl. And it should be transliterated according to Cyrillic <-> Latin rules.
I see. Would it make sense to have the common conversions between ekavian and iyekavian be stored in the conversion tables as well? This way you don't have to use the ((ekavian//iyekavian)) markup all the time.
Also, user should be back to variant which (s)he changed, not to general page (if it is possible, of course).
User can set the preferred variant in the preference. I think when this is finally implemented, most user would just pick one variant and stick to it, and only occationally would want to see how the article looks in other variants.