[conversation] understand images

[rhwy]

when the user send an image we actually don't detect it and just continue with the actual sequence.

prefered behaviour:

• if an attachement of type image is detected in the user input
• we'll search for a random card in "Thank you" intention and show it to the user
• then we'll continue with the next sequence

[rhwy]

done