Article about feedback generation

[filipsch]

As you write chains of SCTs, every function 'contributes' to the feedback message that is generated in case the student made a mistake. Some (not all, like in pythonwhat) of these 'chunks' of feedback can be adjusted by setting the appropriate arguments in check_ functions. An article should be added to the testwhat docs that:

• explains in understandable terms how message generation works
• gives examples about it
• explains how you can customize messages generated by the SCT

Requested by @klmedeiros See similar issue for pythonwhat