Create AI image rendition

When creating ML models, a lot of times, we repeat the image export or converting a big file:

• different models might use the same blob as input (so we convert it for every new model)
• exported file:content does not have picture rendition done
• we don't have the picture facet
• exported a blob that is not file:content
• export failed and we need to export again

Create a new AI rendition that will return a 299x299 resolution image for any blob on the document (ie. files:file, file:content, customer:newBlob, etc..).
On export, if the feature is type `image` we don't recover the blob  + converts, but ask for the specific rendition for that blob.
Outcomes:

• we never repeat a conversion
• we only do the conversion as the rendition is asked (on export)
• the rendition give use the correct dimension for us to export, so no conversion is needed anymore
• we can add more renditions in the future for different types of export (ie. larger image for tagging, 3D if we go there, video if we go there)

(the renditions need to have persistence. Thinking of Picture views where there is a new schema for it on the AI )
(Also, the rendition, in case of picture facet + blob file:content + enough size will still rely on the available picture view to start the conversion from this)