[clang] [KeyInstr] Complex assignment atoms (PR #134638)

[Orlando Cazalet-Hyams via cfe-commits]

OCHyams wrote:

Sorry I misunderstood what you were saying. Fixed, this should now have symmetry with prior patches. Expanded the test coverage. There's 4 binop assign calls that are instrumented individually which reflects the previous patch's macro usage. The expanded test coverage identified a redundant atom group nesting, which is now also fixed (and could only be fixed with the symmetrical approach, hurray). Thanks for the keen eyed review!

https://github.com/llvm/llvm-project/pull/134638