Fix profiling when IO_Fork from the IO package is used

[ChrisJefferson]

Many apologises for the multiple PRs.

After further thought #2907 is both (a) too complicated, and (b) doesn't even do what I needed (it makes too many files). This adds a tiny part of that PR. I will then make a seperate PR to IO, which calls this function when in the child part of an IO_fork.

While this will tie profiling and IO_fork more tightly, I think that's the right option, as the forks in GAP never make processes which can continue to run GAP.

[fingolfin]

This is not a blocking comment, but still: Why the +20? That seems to defeat the purpose of GAP_PATH_MAX. If we are concerned the buffer might not be big enough (a quite reasonable concern, although hopefully in practice rarely of relevance), well, then we should avoid hard coded buffer sizes at all everywhere. OTOH, if we accept that we'll only support paths up to a certain length, then why suddenly make an exception here?

[ChrisJefferson]

It's because otherwise I get a warning from gcc complaining my snprintf could overflow it's buffer (which seems like a stupid warning, as isn't that the point of snprintf?) I couldn't figure out how to disable.

[ChrisJefferson]

However, I accept that's stupid. I've now done it "properly", without having any warnings.

[codecov]

Codecov Report

Merging #2908 into master will decrease coverage by <.01%.
The diff coverage is 10%.

@@            Coverage Diff             @@
##           master    #2908      +/-   ##
==========================================
- Coverage   83.74%   83.74%   -0.01%     
==========================================
  Files         681      680       -1     
  Lines      346547   346275     -272     
==========================================
- Hits       290228   289973     -255     
+ Misses      56319    56302      -17

Impacted Files	Coverage Δ
src/profile.c	43.47% <10%> (-1%)	⬇️
lib/init.g	81.13% <0%> (-0.72%)	⬇️
src/integer.c	97.88% <0%> (-0.18%)	⬇️
src/gap.c	84.34% <0%> (-0.1%)	⬇️
lib/record.gd	100% <0%> (ø)	⬆️
lib/string.gd	100% <0%> (ø)	⬆️
src/julia_gc.c
src/stats.c	95.47% <0%> (+0.19%)	⬆️
src/gasman.c	88.36% <0%> (+0.43%)	⬆️
src/weakptr.c	99.2% <0%> (+0.69%)	⬆️
... and 2 more