Urs Liska wrote
a) in what form should I provide such an image that it can be translated
and modified in the future?
Since your image is simply boxes with text and arrows connecting them, one
possibility might be to do it with html and css rather than making it an
image. The styling and layout of the boxes would be straightforward and the
arrows could be background images. I would think that that would make it
simpler to localize the text.
But I don't know if and how the texidoc generation of html would work for
this.
-Paul