[llvm] [M68k] Fix reverse BTST condition causing opposite failure/success logic (PR #153086)

[Min-Yih Hsu via llvm-commits]

https://github.com/mshockwave commented:

Generally looks good. The key here is that BEQ branches when Z is one, therefore if we're comparing against zero in the setcc we should not flip the condition.
Could you add a codegen test?

https://github.com/llvm/llvm-project/pull/153086