Revert boost upgrade, and update clang to v16

[dagardner-nv]

Description

fixes #379 ?

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[codecov]

Codecov Report

Merging #382 (3999f28) into branch-23.11 (483194d) will decrease coverage by 0.01%.
The diff coverage is n/a.

Additional details and impacted files

@@               Coverage Diff                @@
##           branch-23.11     #382      +/-   ##
================================================
- Coverage         73.11%   73.11%   -0.01%     
================================================
  Files               382      382              
  Lines             13403    13403              
  Branches           1010     1010              
================================================
- Hits               9800     9799       -1     
- Misses             3603     3604       +1     

Flag	Coverage Δ
cpp	69.04% <ø> (-0.01%)	⬇️
py	42.05% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 1 file with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 483194d...3999f28. Read the comment docs.

[cwharris]

Question: What was the reason for upgrading boost and libhwloc? Do we need to re-evaluate those conditions now?

[cwharris]

Boost was upgraded in #364, but I do not see a reason given. Looking through other sources, my understanding is that upgrading boost fixed something related to clang's build. By upgrading clang instead, it sounds like we are bypassing that problem with clang, which gives us the opportunity to use the older version of boost.

We should consider whether upgrading boost in the future is going to be problem, and, more importantly, why it was an issue this time.

[dagardner-nv]

Boost was upgraded in #364, but I do not see a reason given.

The reason was a newer build of clang-15 was uploaded to conda-forge, which broke CI, not just CI actually cmake failed early on compiler check. It wasn't a point release which made it difficult to avoid. The solution then, was to upgrade boost, however that brought about breaking changes in the way promises in boost:fibers are handled.

This pr takes us back to our prior version of boost, and we avoid the bad build of clang-15 by adopting clang-16.

[dagardner-nv]

/merge