Content Model code improvement 3: Refactor List

[JiuqingSong]

I'll have a bunch of PR to improve Content Model code, and do some clean up.

Currently a list item is represented like this:

{
  blockType: 'BlockGroup',
  blockGroupType: 'ListItem',
  format: { ... },
  blocks: [ ... ],
  levels: [{
     listType: 'OL',
     orderedStyleType: 1,
     ...
  }]
}

In the levels section, we mixed list type, list format and list metadata together. In this change. I split the list level into 3 parts:

• list type
• format
• metadata

{
  lisType: 'OL',
  format: { ... },
  metadata: {
     editingInfo: '...'
  }
}

So that we can leverage existing parser and applier to operate on them.

Here is the motivation of this change. With earlier changes, we have split content model dom conversion code and format API/editor code. For most metadata API they are moved into roosterjs-content-model-editor package, only exception is list metadata API. It is still used when we parse and apply list format. But as a DOM conversion code, it should just do simple things -- read whatever list style and then write it back as it is. Other complex work like aggregate list item into list thread, or change list item style, they should be part of the format API code. In order to be able to move those code out, as a prestep we need to refactor the data structure of list item level first.