Fewer duplicate specs for Kernel methods

[Earlopain]

Currently it checks that Kernel.method == Object#method by running specs twice. But the Kernel methods are just mixed into Object, so we should be able to assert that instead.

Draft for now. I think this still asserts the same thing, but do let me know if I should do the same for the other kernel methods/if I'm missing something/it can be asserted in a better way.

[eregon]

This seems a nice simplification, most Kernel methods are just module_function, so both singleton and private instance methods of Kernel.

There seems to be no easy way to prove they are the same, except maybe via their .source_location:

irb(main):001> Kernel.instance_method(:Float) == Kernel.singleton_class.instance_method(:Float)
=> false
irb(main):002> Kernel.instance_method(:Float).source_location
=> ["<internal:kernel>", 194]
irb(main):003> Kernel.instance_method(:Float).source_location == Kernel.singleton_class.instance_method(:Float).source_location
=> true

That's only OK if the returned source_location is non-nil though.
Good enough in this case since it seems the case on Ruby 3.3+, but I guess it won't be the case for most other Kernel methods (e.g. for Kernel#p).
So I wouldn't bother to go this way since it works for a small subset.

I think for Kernel methods it's fine enough to check that there is a private instance method and a public singleton method of the same name, and only test one of them, and assume the other with the same name does the same.

So here we'd test only Kernel#Float (because it's what is most used), and just check that Kernel#Float is a private instance method, and that Kernel.Float is a public singleton method.
And we can do the same for all other module_function methods of Kernel.

[Earlopain]

Turns out that most already only tested a single variant. It's about 700 fewer specs overall, which I assume were mostly caused by sprintf.

I've added private/public checks to all, which is the only observable effect of module_function. I did try (again) to prove they are identical but failed (again). One could also check if arity/parameters/source_location are the identical but I don't think that gives much further value.

[eregon]

Great cleanup, thank you, just a few nits