Hi Samvit!
There are a lot of questions here, so I'll leave a few for the other experts and address efficiency. Here are a few rules of thumb:
70% confidence in your answer in two minutes is better than 90% confidence in your answer in three minutes
Given the time constraints and the adaptive nature of the test, there really aren't any "easy" questions (unless you're scoring 50-51 in math, in which case all the math questions are more or less straightforward and your only concern is sloppy mistakes), so while that extra minute might not seem all that serious the first time, if you have to take it once, you'll probably have to take it again on some other question ... and before you know it, you'll have zero time for four or five nasty questions at the end of the section. Get used to being reasonably confident and moving on: if you really get 7 out of 10 questions right (especially in math), you have an outstanding shot of cracking the 80th percentile in both sections.
Don't check your answers in between questions
Far too many students do this in practice, but it develops an awful habit: the need for affirmation being moving on to the next question (affirmation the real test obviously won't provide). This is hard to unlearn, and if you can't shake it before the exam, you'll waste time pondering how you've done so far. On the test that's useless - what's done is done - so get comfortable with uncertainty.
Don't feel the need to prove (or even find some exact) answers in DS
Data sufficiency is ultimately a test of your mathematical instincts, not of your mathematics: you don't have to prove anything, you just have to be reasonably confident that there's a pattern or that there's only one solution. If I ask you whether x is a perfect square and tell you that x has an odd number of factors, try a few numbers and see if there's a pattern. If you can break it, great - the statement is insufficient. If you can't, and you'll tried some small, funny numbers, then you can be reasonably confident that you're right.
(The key, though, is that if the pattern is going to break, it will break at a rather small, familiar number, like 0 or 1 or 3 or -1/2; it won't be like Elkies' counterexample to Euler's sum of powers conjecture - "So it turns out that your conjecture doesn't hold for 2682440^4 + 15365639^4 + 18796760^4 = 20615673^4." "How foolish of me!")
As an aside, this is why a lot of mathematically inclined students can actually underperform on the math section of the GMAT: they need certainty, and fret (and dally) until they get it. A better way (to give one technique) is to say "C is too easy, and B is missing a piece, so it must be A". It's shocking how often this sort of stuff works. All you have to do is outfox the testwriter, who is subject to all sorts of awful constraints (it has to be one of these five answers, the questions can only test 8th grade public school math, four of the verbal answers to each question must have at least one false/weak word with which to invalidate them, etc.)
Don't spend more than 90 seconds on a SC question for any reason
It isn't worth it. You're just shaking the magic 8-ball, hoping for a clue.
As for IR, I'd exhaust all the officially released questions before moving on to any third party materials: do the 50 that come with OG 13, everything in GMATPrep and its supplemental question pack, and the questions on the mba.com website first.
