Remove class deallocator following deprecation

[Geod24]

See the changelog entry for the timeline.

I wanted to apply the same to class allocator, but @disable new() was not deprecated, with the rationale that aggregates could want to force to use a different allocator.
It sounds very odd that this very specific things could be allowed. There are other ways to do so (e.g. a factory function which accepts a delegate). But that means another deprecation period for them.
CC @JinShil

[dlang-bot]

Thanks for your pull request, @Geod24!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub fetch digger
dub run digger -- build "master + dmd#10727"

[Geod24]

This code was a bit surprising: If ad.aggDelete is false (and it should, because otherwise we error out), the if is always followed and the code that follows is never executed. However the code seemed to account for the possibility of f or fd being null, which doesn't make sense as fd can never be null. Maybe it's just historical baggage.

[JinShil]

I wanted to apply the same to class allocator, but @disable new() was not deprecated, with the rationale that aggregates could want to force to use a different allocator. It sounds very odd that this very specific things could be allowed.

Yes, it is odd. @disable new was not deprecated because it was a technique being employed by at least one project in the test suite. It's possible that the 1 or 2 cases where it was being used no longer exist, so it may be possible to deprecate now, but some users were advocating for it to remain (see #8042 (review)), so I left it in.

This is an unfortunate consequence of not following through on the deprecations -- users will choose to use these odd techniques to workaround limitations in the language, which then requires D to support and maintain the feature, or to go through a disproportionate amount of work to get its removal approved.

[Geod24]

Fix for the Buildkite failure: libmir/mir-random#121

[WalterBright]

It's nice to remove a large chunk of code.

[Geod24]

Green