fix: prevent unnecessary property updating in set_attributes

[baseballyama]

fix: #6751, #6752

This is an alternative way of #6758.

Original PR (#6758) uses dependencies.size before call get_dependencies().
But in my understanding, dependencies have an attribute's dependencies.
But get_dependencies() returns parent node's dependencies.
So below logic looks a little bit weird to me.
(But honestly, I couldn't find any examples of anything being broken by this fix. So maybe this fix is okay...)
(If my understanding is wrong, please correct me.)

const condition = attr.node.dependencies.size > 0
  ? block.renderer.dirty(attr.node.get_dependencies())
  : null;

So I think the original code is okay regarding this part.

---

I think the root cause is different behavior between normal variables and spread syntax.
But I don't know how to fix it.

But I think, this behavior becomes an issue only during user input.
And I think in most cases, attributes are not changed during user input.

Therefore I thought 2 ways to fix this issue.

1. Stop calling unnecessary set_attributes.

The runtime calls set_attributes even if min is not changed but it's not necessary.
So I thought if we can handle it at compile process, this is good.
But associative array (or JSON) can have nested objects.
Therefore compile process will be significantly complicated or impossible to handle.
And I think this is too much at least for this issue.
So I didn't choose this way.

2. Stop setting value if it's not changed (I choose this way in the PR)

This is a simple way.
Based on the assumption that the attributes will not change during user input,
set_attributes will not apply new value if it's not changed.
Actually already attr function did this way.
So It'll just be the same process.

---

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

• It's really useful if your PR references an issue where it is discussed ahead of time. In many cases, features are absent for a reason. For large changes, please create an RFC: https://github.com/sveltejs/rfcs
• Prefix your PR title with [feat], [fix], [chore], or [docs].
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.

Tests

• Run the tests with npm test and lint the project with npm run lint

---

I'm so sorry If @Wojmis3 felt impolite,
And I apologize for the length of the text.

Thank you for reading!🎉

[baseballyama]

fix: #6752

I think this issue is also much similar to #6751.
So I added 1 more commit to this PR.

But still, I don't get a comment from maintainer regarding this issue and PR.
So I'm not sure this fix is reasonable.
But personally, I think this is one of reasonable way.
Therefore I added this commit.

Thank you!

[dummdidumm]

Is there any way we can test this?

[vercel]

@baseballyama is attempting to deploy a commit to the Svelte Team on Vercel.

A member of the Team first needs to authorize it.

[baseballyama]

@dummdidumm

I tried to add a test but I can not make focus on date part of date input.
Therefore I can not test for this.
So in terms of adding a test, I added a comment.

[dummdidumm]

The Chrome date input issue is no longer reproducible, and this PR does not solve the bug with width not updating, therefore closing.