Fix outstanding failures with perl testing

[MikeFultonDev]

Latest version (5.39.5) has the following failures:

Failed 33 tests out of 2599, 98.73% okay.
	../cpan/IPC-Cmd/t/01_IPC-Cmd.t
	../cpan/IPC-Cmd/t/03_run-forked.t
	../cpan/autodie/t/open.t
	../cpan/podlators/t/general/basic.t
	../cpan/podlators/t/text/color.t
	../cpan/version/t/07locale.t
	../dist/Net-Ping/t/001_new.t
	../dist/Net-Ping/t/010_pingecho.t
	../dist/Net-Ping/t/120_udp_inst.t
	../dist/Net-Ping/t/130_tcp_inst.t
	../dist/Net-Ping/t/140_stream_inst.t
	../dist/Net-Ping/t/150_syn_inst.t
	../dist/Net-Ping/t/450_service.t
	../dist/Net-Ping/t/510_ping_udp.t
	../ext/B/t/b.t
	../ext/B/t/f_sort.t
	../ext/POSIX/t/mb.t
	../ext/XS-APItest/t/handy00.t
	../lib/B/Deparse.t
	../lib/locale.t
	../lib/warnings.t
	io/closepid.t
	io/openpid.t
	op/magic.t
	perf/opcount.t
	porting/authors.t
	porting/cmp_version.t
	porting/globvar.t
	porting/regen.t
	re/charset.t
	re/regex_sets.t
	run/locale.t
	uni/overload.t

Previous version had the following failures:

Failed 26 tests out of 2490, 98.96% okay.
	../cpan/IPC-Cmd/t/01_IPC-Cmd.t
	../cpan/IPC-Cmd/t/03_run-forked.t
	../cpan/podlators/t/general/basic.t
	../cpan/podlators/t/text/color.t
	../cpan/version/t/07locale.t
	../dist/Net-Ping/t/001_new.t
	../dist/Net-Ping/t/010_pingecho.t
	../dist/Net-Ping/t/120_udp_inst.t
	../dist/Net-Ping/t/130_tcp_inst.t
	../dist/Net-Ping/t/140_stream_inst.t
	../dist/Net-Ping/t/150_syn_inst.t
	../dist/Net-Ping/t/450_service.t
	../dist/Net-Ping/t/510_ping_udp.t
	../ext/B/t/b.t
	../ext/B/t/f_sort.t
	../ext/POSIX/t/mb.t
	../ext/XS-APItest/t/handy00.t
	../lib/B/Deparse.t
	../lib/locale.t
	../lib/warnings.t
	op/magic.t
	perf/opcount.t
	porting/authors.t
	porting/cmp_version.t
	porting/regen.t
	run/locale.t

The new failures are:

> 	../cpan/autodie/t/open.t
> 	io/closepid.t
> 	io/openpid.t
> 	porting/globvar.t
> 	re/charset.t
> 	re/regex_sets.t
> 	uni/overload.t

Of these tests, 'openpid' received a SIGHUP in the original run (so it should have been listed as 'fail') and io/closepid.t passed.
The other new failing tests did not exist before.

So io/closepid.t may be a regression.

[khwilliamson]

I'm wondering if this can be closed. I've fixed a number of these issues in perl over the years. Today, using

This is perl 5, version 43, subversion 10 (v5.43.10*) built for os390

I get

Failed 7 tests out of 2661, 99.74% okay.
../cpan/ExtUtils-MakeMaker/t/02-xsdynamic.t
../cpan/IPC-Cmd/t/01_IPC-Cmd.t
../cpan/IPC-Cmd/t/03_run-forked.t
../cpan/version/t/07locale.t
porting/cmp_version.t
porting/regen.t
run/locale.t

The porting/ failures are just bookkeeping in the workspace; the locale ones look to me to be because of #73, so are covered by it.. The other 3 are long standing. Maybe they should be filed against perl.