Am I correct getting this impression that what's being said is that auto mechanic work is somehow not "brain work"? Because for solving a lot of car problems it takes brains. If you're good at troubleshooting car problems, you will actually probably be good at programming.
Programming, and especially software architecture, tends to be more abstract thinking.
FWIW, the Kolbe test methodology [1] has a good way of quantifying this sort of natural aptitude. 90% of programmers I know who have taken the test [2] are very good at abstract thinking, and not as good at the physical kind.
[1] kolbe.com
[2] Maybe 2 dozen programmers. So, not a huge sample, but enough to see a trend.
There might be a bit of 3d thinking in figuring out how to, say, measure that in location X of the engine there is current Y, but it's not like you can figure out you have to check the current in that spot without having a clear mental picture of what is going on.
Not to mention remembering that car model X tends to develop fault Y which sometimes causes observable behavior Z, which can be identified by checking A, B, C etc. etc. seems very similar to software engineering to me.