store location of loop boundary

[simeonschaub]

Since lowering always inserts gotos and other instructions after loop
bodies, this allows debuggers to give a more useful location for these
instructions.

fixes JuliaDebug/JuliaInterpreter.jl#485

[simeonschaub]

The failure in Distributed shows this could potentially be an issue for macros, so we should do a PkgEval run.

[simeonschaub]

@nanosoldier runtests(ALL, vs = ":master")

[nanosoldier]

Your package evaluation job has completed - possible new issues were detected. A full report can be found here.

[simeonschaub]

Ok, a lot of the failures do seem to be real. ChainRulesTestUtils and Cthulhu are probably ok, since they are testing Base internals, but Hyperopt and InteractBase seem to have similar issues as Distributed.@distributed.

Are we ok with just fixing those packages, or is this change a bit too risky? Alternatively, we could also make the LineNumberNode part of the for expression itself, but that might be disruptive in a similar way.

[JeffBezanson]

Can you elaborate on why this broke the macro?

[simeonschaub]

The macros extract the loop bodies and actually use the return value of that block expression, which is now nothing because the last entry is a LineNumberNode.

[JeffBezanson]

vtjnash says we should probably fix those packages first, then it might be ok to do this.

[simeonschaub]

OK, all affected packages should now have been fixed. Could I get a review?

[vtjnash]

I don't think this should be backported