[REG 2.072] Revert object.hashOf changes from "use array interface to hashOf()"

[WalterWaldron]

Solves issues 16654 & 16764.

See: #1595 (comment)
Issue links: 16654 and 16764

[dlang-bot]

Fix	Bugzilla	Description
✓	16654	hashOf returns different hashes for the same string value
✓	16764	hashOf is misleading, error-prone, and useless

[WalterBright]

Reverting does not fix the problems #1595 solves.

[MartinNowak]

I think everyone is clear, that object.hashOf is an API for the typesafe and CTFEable core.internal.hash.hashOf, and Walter was just confusing the two (also see https://github.com/dlang/druntime/pull/1595/files#r67473386).
Hard to say anything to the comment above other than, please take enough time to understand the arguments and actual changes of other people, before responding.

[MartinNowak]

Thanks, please change your commit message to
fixes issues 16654 && 16764 (see regex) to close the mentioned Bugzilla issues.
There should be checkmark in the fix column of dlang-bot's comment.

[WalterWaldron]

Thanks, please change your commit message to
fixes issues 16654 && 16764 (see regex) to close the mentioned Bugzilla issues.

Done.

[MartinNowak]

Auto-merge toggled on

[wilzbach]

I also just wasted quite a bit of time tracking this regression down :/
Thanks a lot @WalterWaldron @MartinNowak - you were definitely faster :)

-> anyways: we really need better testing to avoid such major regressions sneaking into releases.