LibGit2/test: Print failed process PID on challenge prompt failure

[Keno]

Makes it easier to locate the failed process in an rr trace after
the fact (cf #45797).

[vtjnash]

It seems like this should be findable by looking for ppid=579 signal=15 in rr ps?

[Keno]

You'd think so, but I don't see any such process, so it may have exited without the parent noticing or something.

[vtjnash]

pid 1978 in julia-9 is the only child that reports to be killed with signal 15

[Keno]

pid 1978 in julia-9 is the only child that reports to be killed with signal 15

Oh, so it does. I missed it.

[Keno]

rr doesn't like that particular pid either: rr-debugger/rr#3285

[DilumAluthge]

@Keno @vtjnash Once CI is all green, is this good to merge?

[staticfloat]

I have an alternative solution, which is to print the PID in jl_critical_error() when we are signalled: #45798

[giordano]

Is this PR still relevant after #45954?

I have an alternative solution, which is to print the PID in jl_critical_error() when we are signalled: #45798

I presume you meant to link #45905 instead of this same PR?

[Keno]

Probably worth doing both. @staticfloat's change relies on the child being alive enough to run it's signals handlers, which it may not be sometimes.

[vtjnash]

May be worth doing still, but heavily conflicted now for a small change. And hasn't seemed necessary lately (just a test enhancement)