Support for intervention/image:^3.4.0

[kevindierkx]

This PR's primary focus is to support intervention/image:^3.4.0, it comes with some breaking changes however.

Text wrapping

Wrapping of long sentences is now handled by intervention/image directly, truncating and the removal of lines of text that wouldn't fit in the containing 'box' is handled by the-og.

This heavily relies on the FontProcessor implementation of the used driver, eg. Imagick or Gd. Since we need to calculate the dimensions of a TextBox, to correctly calculate the 'anchor' point/position of related items, the 'canvas' ie. the ImageInterface has been moved from the feature's render method to an earlier stage using the added setCanvas() method. This is currently placed in the addFeature() method, called during the render cycle of the layout.

This modifies the signature of the render() method and will therefore break custom "Features".

Addition of the dimensions() method

The dimensions() method essentially replaces the getPrerenderedBox() and getRenderedBox() methods. The latter were primarily used to calculate anchors and line lengths. Unless a custom "Feature" would call these directly this shouldn't be as impactful.

Removal of the TextModifier implementation

This class was added at some point to calculate the bounding box of a text block. Since this is now directly handled in either the dimenstions() method or the TextBox implementation this class is now obsolete.

Driver agnostic improvements

Because we now directly interact with the image instance for various operations that would require a driver we don't need to initialise drivers or driver related instances ourself. Except for the PictureBox everything should now also work with GD, not sure if anyone was waiting for GD support... but hey it's there.

CTA placement fixes

Some default layouts were missing the preferred position, these have been added and the CTA element is now correctly displayed for all layouts.

Example output

Using the example from the docs, the image below is now the resulting output.

[simonhamp]

@kevindierkx this looks great but the tests are failing... I think you'll need to refresh the snapshots and commit those changes

[kevindierkx]

@simonhamp I've update the snapshots. The placement of the first two using the avatar seems a little off:

• 0a2f319#diff-99cce31f1e3bc63184123292bfbf98a8de13f439d2ea0bc3d71e09b18d9580b8
• 0a2f319#diff-a129afbd5fdf8d345105e84e8eb853f2e2ac94390c2862ff3f081efa7a952eb9

The other snapshots word-wrap a little differently now when using the intevention wrapping logic, which would be expected.

I'll try to review those avatar layouts next week.

[simonhamp]

@kevindierkx and this one also has something up with the background repeating where it didn't repeat before