feat(ui): img2img UX

[psychedelicious]

Summary

Changes to reduce img2img friction:

• Add fit modes to transform:
  • fill: entity is stretched to fill the bbox entirely (current behaviour)
  • contain (new default): entity is scaled up to fill the bbox, retaining aspect ratio, centering the transformed entity in the bbox
  • cover: entity is scaled up so that the smallest dimension fills the bbox completely, centering the transformed entity in the bbox
• Some prep to make room for a new layer context menu item:
  • Create IconMenuItem component (revised abstraction of the icons we have in a row on the image context menu, fixed an issue where the buttons cannot be disabled)
  • Use the new component for these layer menu items: 4x arrange actions, duplicate, delete
  • One kinda awkward thing, the Reference Image context menu now only has a row of 6 icon buttons...
• Add ability to ignore the cache when rasterizing a layer, fixing an edge case where transform would do nothing
• Disable the transformer controls (i.e. the drag handles) while applying the transform (i.e. rasterizing and uploading)
• Add silent option to transformer, which prevents any UI or controls from being shown while transforming
• Add layer context menu item for Fit to Bbox, which uses a silent transform to immediately fit the entity to the bbox using the contain mode with a single click

Related Issues / Discussions

Discord & offline discussions about img2img friction

QA Instructions

Try the new options out:

• Transform a layer and try the fit options
• Try Fit to Bbox option in layer context menu
• Review the new image icon buttons

Merge Plan

n/a

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Tests added / updated (if applicable)
• Documentation added / updated (if applicable)