From what I can gather:
hyper-extending is when the elbow moves backwards, past the angle of 180 degrees (i.e., the "non-natural" way, if that makes sense).
adduct and
abduct refer to sideways movement of the elbow - I presume one is "inside", and the other "outside", if that makes sense.
The conclusion that Angus Fraser and the rest of the team given the task to look at these issues seem to have come to is that they will not penalize those (like Shoaib) who hyper-extend, abduct or adduct, because they consider them completely involuntary movements, caused by the force of the arm as it rolls over during delivery. So the tolerance levels do NOT pertain to this. They refer to the usual concept of the straightening of the elbow in a throwing action, like you would return a ball in the field.
The reason 15 degrees was chosen was not completely arbitary. According to the study, it's the degree to which an element of excessive straightening is detectable to the naked eye. There also seems to be an assumption that once we get to this level of straightening, there's room for voluntary remedial correction (because the player has some control over that degree, as opposed to the hyper-extending, etc, etc).
If you want more stuff apart from the cricinfo articles, Angus Fraser's articles at the Independent are a good place to start...
http://sport.independent.co.uk/cricket/story.jsp?story=581257
http://sport.independent.co.uk/cricket/story.jsp?story=581256
Given that the existing rules were based on the degree of orthodox (for want of a better word) straightening, and NOT hyper-extending, adduction, abduction, etc, I think your questions are resolved, and that it does seem that Murali is no worse than the majority of bowlers out there, past and present. In my opinion, anyway.
I think the only refutation available for those that still want to isolate Murali is that those studying this issue are lying.