Block Library: Implement pass-through behavior for Column block

[aduth]

Related: #15927
Closes #7694
Partially addresses #15660

This pull request seeks to implement a "pass-through" behavior for blocks.

It is implemented as two parts:

• Anticipate wrapper props which disable a block's ability to be focused by bypassing the default selection behavior (allow the event to propagate to an assumed ancestor).
• Assign as a prop of the Column block wrapper to disable focus in Column block.

With these changes, it should become impossible to select a Column block either by clicking on it, or by using arrow keys to navigate blocks. It is technically still possible to select a Column block via the Block Navigation menu. I don't personally have an issue with this, since the block is technically still present, and the primary motivation is in solving the usability problems around standard selection interactions (which this achieves).

Blocked: While this pull request is considered complete so far as implementing the pass-through behavior, it would need to be merged only in combination with a separate pull request which would recreate the existing Column attributes customizations of widths and vertical alignment assignment. My plan is to create separate pull requests for these distinct efforts, and merge them all at once only when they have each been approved.

Testing Instructions:

Verify that click and arrow key navigation to a Column block is not possible, but that otherwise behaviors are unaffected:

• Click-through behavior implemented in Selecting Parent Blocks: Try clickthrough #15537
• Columns selection
• Selection of blocks within a Column
• "Focus" selection of various block types (e.g. Paragraph, Image, Code blocks each have distinct selection behaviors worth validating)

[gziolo]

There is one line which needs to be double checked. Otherwise, code looks good and tests well.

[gziolo]

I noticed that there are 2 failing e2e tests. They both seem to be related to the changes introduced.

[aduth]

I noticed that there are 2 failing e2e tests. They both seem to be related to the changes introduced.

Yeah, while I've not yet looked at the specific failings, in reflection I recall there being some test cases written as assuming the column as a "stop" point for, e.g. ArrowDown. These would need to be updated since that's no longer true.

[aduth]

We may revisit this in the future, but for now this is not mergeable in its current state.