[RFC] Stabilize the useInput return properties

[Nickman87]

I was writing some more complex custom components inside of react-admin that use useInput to update fields.
Somewhere in my code I use an effect, based on the results of this hook, to trigger updates like so:

const [someState] = useState("someValue");
const testField = useInput({source: "test"});

useEffect(() => {
  if (testField.field.value !== someState) {
    testField.field.onChange(someState);
  }
}, [someState, testField.field.value, testField.field.onChange]);

(I know this might look like stupid code, but this is a very trimmed down example of what I want to achieve in my component:) )

This resulted in an infinite loop because the onChange that is returned from useInput is not stable, it is always a new function, even though nothing significant changes.

I made a quick fix for this in this PR, will add tests or more fixes if this is deemed to be a good fix or not. Not sure if there are further implications or reasonings why this is done the way it is now, so looking for feedback.

[slax57]

I believe that's because we never had a use-case where this would be a problem, but currently I see no good reason not to memoize the onChange and onBlur callbacks.
I'm just not a huge fan of the name stableOnChange. You should just call them onChange and onBlur IMO.
But that's definitely a PR I would approve if it doesn't break anything!
Thanks for submitting it!

[fzaninotto]

I made another PR that should solve the problem without adding new props. Thanks for bringing attention to this problem!