Clean up numeric EditorProperty setup() methods

[lodetrick]

Currently, the numeric EditorProperty setup() methods are somewhat brittle. When these properties are created, they are first bundled into a struct, which is then destructed in the method call. This PR changes the signature of the numeric setup() calls to use the struct directly. This has several benefits:

• If any new functionality is wanted for these properties (like in Allow @export_range to support numeric types #105878), this change makes it easy to make these changes without adding any arguments.
• It removes the use of "magic numbers". In several internal files, EditorPropertyIntegers are created directly (I don't know how much this should be encouraged). Their setup() calls have many seemingly arbitrary numbers and booleans that are hard to understand without looking at the editor_properties.h file. This change makes it clear within the file what the parameters that the code is changing.

Note: In making this change I found multiple invalid calls to setup(), most looking like: setup(-1, 99999, true); where there is a boolean in the third argument (which is defined as double p_step). I'm not sure what the intention was originally, but making this change will prevent future invalid calls and help catch other invalid calls before they are propagated.

[KoBeWi]

This could wait for #100749
C++20 has a nice syntax for initializing structs, like

property->setup(EditorPropertyRangeHint{ .min = range.x, .max = range.y, .or_greater = false, or_less = false });

etc.

[KoBeWi]

Looks good.

The new syntax I mentioned in #108065 (comment) could be applied later, the struct doesn't look directly created that much.

[lodetrick]

Updated the function signatures as suggested by KoBeWi

[Repiteo]

Thanks!