Fatalise swapping a 'use VERSION' declaration, as previously promised

[leonerd]

Around 5.40 we said this would become fatal in 5.44. Since we're nearing 5.44 completion we should actually make good on this.

• This set of changes requires a perldelta entry, and I need help writing it - I don't know where in the file it should be added

[ap]

It’s too late to do this in 5.44. For now we need a PR to change the warning to say 5.46 and then we need to revisit this as soon as the next dev cycle starts.

[jkeenan]

Around 5.40 we said this would become fatal in 5.44. Since we're nearing 5.44 completion we should actually make good on this.

* This set of changes requires a perldelta entry, and I need help writing it - I don't know where in the file it should be added

This was originally discussed in #23624, filed last August. As I reported there, I created a branch in which I encountered many test failures, including some which your p.r. is encountering. I was able to fix some but not all of those failures, so I never promoted my branch into a pull request. AFAICT no one else picked up that branch. As I indicated in #23624 (comment) three days ago, "Promotion of a deprecation warning into a fatalization almost always entails CPANtesters failures, which meets the minimum criterion for a "controversial change." Since we're past the "controversial changes" release deadline, this work will have to be deferred" to the next dev cycle. But in the meantime I encourage you to pick up the branch I was previously working on.

[leonerd]

I've opened #24289 to just change the scheduled version number.

[tonycoz]

The change looks fine, obviously deferred.