Element: Avoid double-escaping valid character references

[aduth]

This pull request seeks to improve the element serializer to avoid double ampersand encoding of valid character references.

Before	After
serialize( 'Foo & Bar & Baz' ); // "Foo & Bar & Baz"	serialize( 'Foo & Bar & Baz' ); // "Foo & Bar & Baz"

Testing instructions:

Ensure unit tests pass:

npm run test

Verify that unexpected double-escaping no longer invalidates blocks:

1. Log in as a contributor
2. Navigate to Posts > Add New
3. Insert a paragraph block with some text
4. In Blocks > Advanced, set "Additional class names" to "foo&bar"
5. Save the post
6. Refresh the page
7. Note the block is not invalid

[noisysocks]

Very cool! 🎉

[noisysocks]

When I read 'imperfect' it makes me think that the function is unfinished, when in fact it's intentionally imperfect.

[aduth]

When I read 'imperfect' it makes me think that the function is unfinished, when in fact it's intentionally imperfect.

Depends how you view it. Technically it could be improved, but as I saw it the only option would be to include a full list of valid named character references, and in weighing the cost in doing so, I thought it to not have a net positive benefit.

[aduth]

Thanks for the review @noisysocks ! I'll plan to merge this sometime after the upcoming 3.0 release.

[gziolo]

It looks like this PR was missed, 3.1 is coming up soon, let's get it in :)

[aduth]

Thanks @gziolo , I carelessly forgot about the pull request, but fortunately at least included it in the 3.1 milestone so it didn't go overlooked 😄