c47db038 was an attempt to fix fdo#52547,
i.e. to make sure that the actual image data was stored if an image was
being pasted from a browser (where the clipboard contains both image data,
but also an html construct which would instead embed the image from the
web). However the new ordering inadvertently meant that strings would
be preferred to html -- i.e. copying a mixture of text+image from a
webpage would result in only the text being stored.

The new ordering causes images copied from the browser to still have
the actual image data stored (rather than a link in the case of the
html being stored), but html is still preferred to plaint text.

(This is primarily as a result of Firefox (and possibly other browsers)
changing their clipboard to contain an html construct which links to the
original image, in addition to the raw image data-- and html previously
being preferred to images in the paste ordering meant that the linked
version would be pasted by default, which is not the expected result
when copying an image.)

Change-Id: Ice9b37cf3dd25d8cdb0e04f19a6b2f3661018564