To answer your first question:
There is no "reasonable" CTR, and as you say, it varies from site to site. The only relevance to you is what happens when you change something - colours, banner size, banner position etc.
But you really have to let your experiments run for a while before you can form a reasonable opinion on whether (a) is better than (b)
I customised the colours in my Adsense ads on one of my sites. I've let it run for 10 days before deciding that I ought to change, because I now think that it adversely affected the CTR. It went down from 'x' to '0.8x' (see, that's how we can discuss CTRs!)
In the 1st full day since changing, the CTR has shot up. But I can't draw any conclusions from this. Again, I'll wait at least 10 days before I can form a judgement.
Even then, there's all sorts of other factors that can intervene.
Hope this helps in some way.