That was a cool breakdown of your thought process. Let me see if I can reinforce a few things and clarify a few things.
You were right in thinking that the presence of a "Thus" and a "so" suggests that there is a main conclusion and a subsidiary conclusion.
To figure out which is which, try the Therefore test (to identify which idea supports the other).
Does 2 support 4?
During a recession, the air pollution from car exhaust decreases, THEREFORE cars emitting pollutants are used less.
Or does 4 support 2?
Cars emitting pollutants are used less, THEREFORE, the air pollution from car exhaust decreases.
This 2nd ordering makes more logical sense, so 2 is the Main Conc and 4 is the subsidiary.
The argument would break down like this:
During a recession, the air pollution from car exhaust decreases.
Why?
Because cars emitting pollutants are used less.
Why?
Because during a recession unemployment rises and so fewer people are commuting in cars to jobs.
If we want to play devil's advocate to this Main Conclusion, we need to think, "How can I make the counterargument that car exhaust pollution does NOT decrease?"
We have to accept that fewer people are commuting to work in cars, but is there any other way that they could be contributing just as much (or more) exhaust as before?
Maybe they now need to drive to job interviews, or the welfare office, or they just drive with vacant stares wondering what became of the beautiful life they once knew.

Before we dive into the answers, let me remind you what a Sufficient Assumption means. You seemed to be using the concept to mean "something that's stronger than what we MUST assume". But a Sufficient Assumption is more than just that; to qualify as a Sufficient Assumption, the idea must logically GUARANTEE the truth of the conclusion.
It will normally sound boring, mathematical, mechanical.
Here, a Sufficient Assumption would be:
"Any time fewer people are commuting to work, there is less air pollution from car exhaust."
It would have to take something we know to be true from the Premises and link it to the wording of the Conclusion, GUARANTEEING that the conclusion is true.
So none of these trap answers are sufficient assumptions. (None of them prove 'there is less air pollution from car exhaust').
====answer choices===
A) "People who have never been employed" seems out of scope. The author was talking about "People who WERE employed" ... he thought that the reduction in air pollution was coming from people who WERE employed but now are not. His argument doesn't address people who never were employed.
If we negate (A), we get "people who have never been employed drive LESS frequently during a recession". Does that hurt the argument?
No. The author is trying to prove that there's less car exhaust during a recession. When we negate (A), it STRENGTHENS the idea that there is less car exhaust during a recession.
B) "most", as you awesomely indicated, is a dead giveaway on Necessary Assumption answers. Out of all the Nec. Assump. answers that have ever included the word "most", perhaps 1% have been right. It doesn't matter to the author's argument whether 51% or 49% of air pollution comes from car exhaust. Whatever the percentage of total air pollution car exhaust represents, the author is arguing that it gets lower during a recession.
C) Again, "most" is a huge red flag that ultimately kills this answer. It's not relevant whether 51% or 49% of employed people use public transportation or not.
Interestingly, though, it IS relevant whether 100% or less than 100% of ppl use public transportation.
The author is arguing that people who lose their jobs will no longer be commuting in their cars, thereby decreasing exhaust.
Well, if EVERYONE was taking public transportation, then there wouldn't be any cars being taken off the road.
So if (C) said "at least some people who are employed do not use public transportation to commute to their jobs", it would be a necessary assumption.
D) When we negate this answer, it says that "decreased use of cars due to not having a job IS offset by increased use of cars for other reasons". Boom goes the dynamite -- this kills the argument.
E) The author doesn't have to worry about whether car drivers are more likely to lose their jobs than non-car drivers are. He just needs there to be at least some car drivers who lose their job and end up driving less as a result.
Hope this helps.