Fix UTF-8 character decoding error on RST parsing

Currently, if the user types a UTF-8 character on a database documentation (full-length description field), the system generates a 500 error trying to parse it. It is harmless in the sense the object is not created nor the system gets into any funny state, just annoying.

Here is some bogus document to try:

In CASMEⅡ, we provide 5 classes of micro-expressions

We should protect the relevant pages.