remove Core._apply Builtin

[vtjnash]

We can emulate this deprecated function, until we delete it in v2.
This in turn revealed some missing error checks, which we also add.

Fixes #39113

[maleadt]

@nanosoldier runtests(ALL, vs = ":master")

[nanosoldier]

Your package evaluation job has completed - possible new issues were detected. A full report can be found here. cc @maleadt

[StefanKarpinski]

This seems like it should be removable: it's not exported and it starts with _; how much more internal can a function be? It would be good to fix any packages that removing it breaks, but deleting is definitely not disallowed by semver.

[simeonschaub]

Should this also be backported then?

[simeonschaub]

This seems like it should be removable: it's not exported and it starts with _; how much more internal can a function be? It would be good to fix any packages that removing it breaks, but deleting is definitely not disallowed by semver.

While it might not be covered by semver, there are still a lot of packages using it: https://juliahub.com/ui/CodeSearch?q=_apply&u=all&t=all. I don't see any issues with just keeping it for 1.x, so I don't think there's reason enough to break a bunch of code by removing it.

[Keno]

Yeah, we can't remove it. This fallback might be ok though.

[StefanKarpinski]

Why can't we remove it?

[Keno]

Because it would be extremely breaking for no reason. This used to be inserted by the frontend, so it has propagated far and wide. It's also inside things like serialized closures inside Flux model files.

[vtjnash]

Looks great. The only failure this causes is Zygote, where we fixed a test_broken. GTG.

[KristofferC]

Doesn't backport cleanly and my attempts at fixing it still errors. Can push a rebased commit to #39160.