I am a "rules" guy too. Evidence: read all the conditional statement and necc/suff questions I've asked.
But as you pointed out, the LSAT, at least at the higher level, encourages flexibility and pattern recognition. To this end, I believe tools like the "scale" for RC, and the "logic chain" for LG binary grouping games are just tools. I think we do ourselves a disservice when we try to fit the LSAT questions exactly to the tools -- it should be the other way round.
To answer your question, I use the idea of the scale, but not always. For example, if the RC passage reads like a descriptive essay, I don't actively look for opposing sides. Sometimes I'll even go through the RC passage without clearly mapping out the sides, but I can still attack the questions with knowledge about the general structure, main point and gist of support.
Of course each person has his/her own style. However, the reason I posed the question above is because I think a test taker is more likely to do better if they don't approach RC questions with the mindset of "Gosh, I couldn't find the scale -- I must not understand the passage!" -- Well, maybe there was no scale (opposing sides) to begin with