add updating guard to binding callback

[tanhauhau]

Fixes #3180
Fixes #5117

binding callback should be called only when changes happen internally.

similar with <input bind:value>, the on input only call when typing into the input, but not when programmatically set the input.value.

a brief description of the bug:

• we have a <Component bind:value /> and when we $$invalidate(value) from outside,
• component.$set({ value }) which internally calls $$invalidate(value_internal)
• which then calls the binding callback $$.bound[value]()
• in the binding callback, we call the $$invalidate(value)
• which again calls the component.$set({ value }) which internally calls $$invalidate(value_internal)
• only this time round, the !safe_not_equal(value, value_internal), so the binding callback, $$.bound[value](), did not get called.

as you can see from the sequence of steps, $$invalidate(value) get called twice.

Also, it seemed like it will add quite some bytes into the compiled code, should we rename the variable to a shorter name?
or use 1 / 0 instead of true / false?

Before submitting the PR, please make sure you do the following

• It's really useful if your PR relates to an outstanding issue, so please reference it in your PR, or create an explanatory one for discussion. In many cases, features are absent for a reason.
• This message body should clearly illustrate what problems it solves. If there are related issues, remember to reference them.
• Ideally, include a test that fails without this PR but passes with it. PRs will only be merged once they pass CI. (Remember to npm run lint!)

Tests

• Run the tests with npm test or yarn test)

[Conduitry]

The test failure here appears to be legitimate. I'm marking this as Draft for now.

[Conduitry]

Hrm. I'm not thrilled about the per-component byte cost that this adds. I don't have a better suggestion right now.

[Conduitry]

What if we rework this a bit so that $set() is defined on the class itself, where it sets and unsets the guard flag, and calls something like this.$$set() or whatever, and it's that function that each instance overrides? Still a slight increase in cost, but less of a per-component one than this.

Actually, if we're doing that, maybe it makes sense to redo the change in #5123 slightly so that the is_empty check happens in this new shared $set() method.