Add comment and assertion to pidfd_open workaround

[jakkdl]

EDIT: assumptions were wrong

Ended up researching this during discussion in #2792, and thought I might as well add a comment (and an assertion to double-check assumptions). Also cleared up a coverage miss for an error that shouldn't trigger in test runs.

One could possibly raise a warning instead of the assertion to be extra defensive, but afaict it really shouldn't trigger on 3.9+ so the assertion+message is already very defensive IMO.

https://docs.python.org/3/library/os.html#os.pidfd_open

[codecov]

Codecov Report

Merging #2799 (4f17c57) into master (c16003f) will increase coverage by 0.01%.
The diff coverage is 100.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2799      +/-   ##
==========================================
+ Coverage   98.95%   98.97%   +0.01%     
==========================================
  Files         115      115              
  Lines       17011    17009       -2     
  Branches     3045     3044       -1     
==========================================
+ Hits        16833    16834       +1     
+ Misses        123      121       -2     
+ Partials       55       54       -1     

Files Changed	Coverage Δ
trio/_subprocess.py	98.29% <100.00%> (+1.25%)	⬆️

[jakkdl]

Ah, turns out pypy doesn't have os.pidfd_open

[A5rocks]

-snip-

Never mind, this looks good. I forgot that our workaround will also only be available on Linux >=5.3 cause that's when the syscall is available