Instrumented `$foo= bar;` doesn't need to update `$foo` variable

[Conduitry]

$foo = bar; currently results in set_store_value(foo, $foo = bar);, but this $foo = is unnecessary, as set_store_value will result in the relevant subscription being synchronously called anyway. The $foo = in the compiled code is even a bit misleading in cases (such as tweened and spring stores) where calling .set() on the store doesn't actually immediately update it to that value.

[Conduitry]

I'm upgrading this to a bug, though admittedly an obscure one, because there is no guarantee (nor do I think there should be one) that calling .set() on a store will in fact synchronously set it to anything.

[Conduitry]

This is a bit more fiddly with stuff like $foo += 42; or $foo++;, where we can't just grab the right hand side of an assignment for the second argument to set_store_value(). Alas.

[Conduitry]

There's even more I wasn't thinking about: The assignment could be a destructuring, which I don't see a nice way of handling right now without first assigning the $-prefixed autosubscription variables. Ehh. I'm putting this down for the time being. If I have a situation where I need a store update to cause nothing to happen synchronously, I'll just use foo.set() and not $foo = .

[Rich-Harris]

Now that Svelte 5 is de-emphasizing stores in favour of state, I think we can probably just file this under 'quirks you don't really need to worry about any more'