Run tests in parallel by default

[NickSdot]

Extracted from #22917.

Runs the test suite in parallel by default; automatically selects up to 10 workers. Explicit -j1 remains the opt-out to run tests serial. Smaller batches improve balancing.

Intended to merge after#22957 so eligible Windows test subprocesses no longer use the shell wrapper when parallel execution becomes the default.

Result

Values in seconds; measured on macOS.

mode	master	this PR	saved	speedup
(none)	2223.224	408.009	1815.215	5.45×

Reproduction:

sapi/cli/php run-tests.php -q

Ref: 9b36b91 (review comments)
cc @arnaud-lb

[LamentXU123]

Not an expert in conflicts file. Should cc @bukka in this.

[NickSdot]

Not an expert in conflicts file. Should cc @bukka in this.

FWIW, the doc change is only a clarification because after this tests run in parallel by default. This PR has no changes to CONFLICTS itself. 🫡

[arnaud-lb]

Looks good to me otherwise, but I will wait for other reviews

[arnaud-lb]

Looks good to me otherwise!

I will wait a few days and merge if there are no objections.

[NickSdot]

Looks good to me otherwise!

I will wait a few days and merge if there are no objections.

Friendly reminder, @arnaud-lb. 🫡

If this is merged #22995 would have a slight change. The same is true for #22946, so if you would merge here it would be nice if you could look at the Windows one-liner along with it. I plan to later today push the update for the CI PR.

Monday is a good day to merge, isn't it? 🤓 Will be around to have an eye on CI in case anything unexpected happens!

Edit: does this need a NEWS or INTERNALS entry?

[arnaud-lb]

Yes, this is worth a NEWS and INTERNALS entry. Please add them, and I will merge :)

[NickSdot]

Yes, this is worth a NEWS and INTERNALS entry. Please add them, and I will merge :)

That's done! 🫡

[arnaud-lb]

Thank you!

[remicollet]

Notice, this also affects extensions (pie, pecl...)

So some relying on default behavior need a fix (ex: yar PR #189, memcached PR #581, and probably others)

[NickSdot]

Notice, this also affects extensions (pie, pecl...)

So some relying on default behavior need a fix (ex: yar PR #189, memcached PR #581, and probably others)

@arnaud-lb is this something I should look into or is this acceptable for 8.6 and hence something extensions should take care of (or simply run with -j1)? I'd assume it's fine?

@remicollet what's your take? Like, running with j1 brings back the old behaviour - isn't that fine and extensions can just catch up with flakes (as we do here)? Or do you have something else in mind?

[arnaud-lb]

As this is documented in UPGRADING.INTERNALS this seems fine to me, but maybe @remicollet has another opinion?