Re-sync extension/pytree/function_ref.h with LLVM

[swolchok]

We had an old version, and that version didn't work quite right when
handed a const ref to a callable. Just resync and note divergences
from the original in the comment. Our old version also had an
explicit constructor, which is not desirable because it
complicates callsites. It was also a divergence from both LLVM and
the upcoming C++26
std::function_ref.

The test made invalid use of a FunctionRef -- you can't keep one
around past the lifetime of the function object it refers to, but the
test held onto a temporary capturing lambda in an object. I fixed the
test.

[swolchok]

Stack from ghstack (oldest at bottom):

• Adopt runtime::FunctionRef in thread_parallel_interface.h and thread_parallel.h #10442
• Move fully-featured FunctionRef from extension/pytree to ExecuTorch core #10441
• -> Re-sync extension/pytree/function_ref.h with LLVM #10440

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/10440

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures

As of commit 773b183 with merge base 1432243 ():

NEW FAILURES - The following jobs have failed:

• trunk / test-llama-runner-linux (fp32, portable, linux.2xlarge, executorch-ubuntu-22.04-clang12) / linux-job (gh)
  The process '/usr/bin/git' failed with exit code 255
• trunk / test-qnn-model (fp32, w2l) / linux-job (gh)
  Missing file at path: /home/ec2-user/actions-runner/_work/_temp/_runner_file_commands/set_output_4920c4ee-8c84-4855-afb6-1fcc83a93171

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[digantdesai]

s/ref/ref2 perhaps?

[swolchok]

bypassing two apparent tests flakes and merging.