Should annotations of images supplied by a form be presented as a file picker + canvas?

[eyelidlessness]

The current behavior of image annotations with binary defaults is presented as a combination of:

1. The base file upload field
2. The base canvas field as used in the signature widget
3. The form's binary default image inside the canvas

My understanding from discussion with @lognaturel is that the intent:

• from a spec perspective is that the default may be overridden with any arbitrary image
• from a form definition and submission UX perspective is to annotate the specific image provided by the form

Those can't really be reconciled, but what should be done?