[llvm] [TRE] Adjust function entry count (PR #143987)

[Joel E. Denny via llvm-commits]

jdenny-ornl wrote:

> This patch enables this for instrumented profiling cases because, there, we know the function entry count captured entries before TRE.

Does instrumentation happen before TRE and so profiling then produces counts that make sense for the CFG before TRE?  Or is TRE suppressed somehow during the instrumentation compile?

> A bit harder to assert that for sampling-based.

For the sampling case, is this PR's fix just not necessary, or we do not have an appropriate fix and so there are corrupt counts?

https://github.com/llvm/llvm-project/pull/143987