romio: fix MPIU_external32_buffer_setup

[hzhou]

Pull Request Description

It should allocate a buffer size of true_lb + true_extent rather than
just true_extent.

Fixes #5885

[skip warnings]

Author Checklist

• Provide Description
  Particularly focus on why, not what. Reference background, issues, test failures, xfail entries, etc.
• Commits Follow Good Practice
  Commits are self-contained and do not do two things at once.
  Commit message is of the form: module: short description
  Commit message explains what's in the commit.
• Passes All Tests
  Whitespace checker. Warnings test. Additional tests via comments.
• Contribution Agreement
  For non-Argonne authors, check contribution agreement.
  If necessary, request an explicit comment from your companies PR approval manager.

[ashwinraghu]

This looks promising. I've run some of the configurations of the benchmark that used to consistently fail and they all have passed with this change. A valgrind check also has come out clean w.r.t invalid writes.
I will run some multi-node tests.

[ashwinraghu]

Multi-node tests look good as well.

[ashwinraghu]

I am a little puzzled as to why 4.0.1 doesn't seem to hit this issue because MPIU_datatype_full_size() looks identical there too.

[hzhou]

I am a little puzzled as to why 4.0.1 doesn't seem to hit this issue because MPIU_datatype_full_size() looks identical there too.

I just checked with current main branch, the same invalid writes show up in valgrind without this patch. Segfault sometimes does not trigger due to random runtime memory usage.

[hzhou]

test:mpich/ch3/most
test:mpich/ch4/most
✔️

[ashwinraghu]

Looks good.