Fix Python 2.5.6 for parallel pgen invocation

[nanonyme]

Backported from https://github.com/python/cpython/commit/9828b56347f6171a532dce28c9dc5ab9bd2c8139.patch
Fixes #321

[nanonyme]

@fosslinux looks like build is failing as part of pkg-config removal. I'm running a pipelien on our side at https://gitlab.com/freedesktop-sdk/freedesktop-sdk-binary-seed/-/jobs/5467602634

[nanonyme]

We are mirroring our sources so the removal doesn't really impact us. We are using method for obtaining sources where we prefetch them with manifest since we anyway do not actually boot a system but run something similar to Bubblewrap.

[nanonyme]

Unfortunately it looks like this patch changes hashes. I will need to find out if they are changed in a reproducible way.

[fosslinux]

Yes, the pkg-config thing is expected to fail builds at the moment.

Many things can change hashes, I'm not particularly concerned that hashes are changed - just that the new hash is always the same :)

[nanonyme]

Does the pkg-config situation effectively mean nothing can be merged for the time being?

[fosslinux]

The CI should still pass, because it's using download-distfiles.sh (external distfiles)...

Things can still be merged.

[nanonyme]

This CI ironically would most likely be considerably faster with parallel build on.

[nanonyme]

@fosslinux it seems this new hash reproduced multiple times so I'm hopeful that it persists. I suspect there is one extra file (the timestamp file) as a result of the patch.

[fosslinux]

The CI already runs with --cores 2, which is what the runners have available.. https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources

[fosslinux]

LGTM