[llvm] [RISCV] Move RISCVIndirectBranchTracking before Branch Relaxation (PR #139993)

[Sam Elliott via llvm-commits]

https://github.com/lenary commented:

I'm for this in principle, based on your reasoning.

Do you have any kind of testcase you could add to show branch relaxation getting something wrong?

https://github.com/llvm/llvm-project/pull/139993