I/6179: Fix alignment property converter

[jodator]

Suggested merge commit message (convention)

Fix: Table's alignment attribute converter should override Alignement feature converter. Closes ckeditor/ckeditor5#6179.

---

Additional information

• The where to convert alignment attribute is a different topic: How the table alignment attribute should be downcasted ckeditor5#6197.

[coveralls]

Coverage remained the same at 100.0% when pulling 2b3511c on i/6179 into be0d759 on master.

[jodator]

There are some changes in the Chrome 80 in comparison to Chrome 79... I've just updated to Chrome 80 and indeed this one test fail. Was green on Chrome 79.

[Reinmar]

👍 for consuming the attribute – that's definitely a good change.

👎 for changing the priority of the converter. As I commented in ckeditor/ckeditor5#6179 (comment) the alignment feature converter should be smarter. Semantically, the alignment feature applies to text blocks (it sets text-alignment so it affects the content of those elements). So, it should not apply to object elements.

In other words – IMO, instead of rising the priority of the table alignment converter we should make the alignment feature's converter apply only to non-object elements.

[jodator]

-1 for changing the priority of the converter. As I commented in ckeditor/ckeditor5#6179 (comment) the alignment feature converter should be smarter. Semantically, the alignment feature applies to text blocks (it sets text-alignment so it affects the content of those elements). So, it should not apply to object elements.

In other words – IMO, instead of rising the priority of the table alignment converter we should make the alignment feature's converter apply only to non-object elements.

As in F2F conversation: re-writing the Alignment feature converter is not a half-hour task so we extracted this to: ckeditor/ckeditor5#6199.

@Reinmar other than that this PR is ready to review (and merge - the only failing test fails on master also because of failing tests on Chrome 80.

[jodator]

Invalidated by #234 - the tests remained untouched though.