reverse stacktraces (again)

[pfitzseb]

Fixes https://github.com/JunoLab/atom-julia-client/issues/374.

I'm strongly in favour of this just to be consistent -- is this okay with you, @MikeInnes?

[MikeInnes]

Well we could also be consistent, at least within Juno, by changing the REPL stack trace order.

I can go along with this if there's a general consensus that consistency is more important here, but I really do think it's the wrong default for 99% of Julia users (especially the ones using Juno), who usually have an issue in their code, not at the deepest level of the base system.

[aviatesk]

I'm actually very in favour of @pfitzseb . I'm just an inexperienced Julia programmer, but feel like we beginners want to first refer to the outermost call and then continue to dig into inner calls since it would be easier to start thinking from higher level of the error (assuming people usually look at errors from up to down).