Jordan Rome

Closing this. We can discuss implementation details on the PR.

@danobi I had to add more tests to aot_skip list because AOT seems to have an issue with tuple printing.

> Why change this? Semantically they are identical but adding () adds more visual noise. On top of that, this is no longer backwards compatible for anyone parsing output. It...

> In my recent AST refactorings I've moved us away from representing empty lists with null pointers in an attempt to simplify the code. An empty ExpressionList is now just...

@viktormalik Thanks for the review! I'll wait at least a day or two before merging to give other folks a chance to voice concerns.

> What about JSON output? My preference would be backwards compatibility, supporting multiple versions of bpftrace is a hassle if one parser has to handle different output formats. @janca-ucdavis Right...

> Is there any performance delta? I'm using 5-dimensional maps now and expect to go to 7 soon. There shouldn't be. I think the only difference is a slight increase...

- Fixed the issue with this and added a semantic_analyser test for it: `bpftrace -e 'BEGIN { $a = (3, "hello"); @m[$a] = 1; $a = (1,"aaaaaaaaaa"); @m[$a] = 2;...

@ajor I think I finally addressed all the concerns 🤞

@danobi - Thanks for the review! I made the updates. > I got a bit lost in semantic analysis and codegen changes. hard to tell apart refactor changes with actual...