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.