Unicode normalization

[agitator]

After further discussion of these issues
plone/Products.CMFPlone#2912
zopefoundation/Zope#670
we cam to the conclusion, that the best place to fix this would be normalize any characters in fields that handle unicode.

Proposed solution:

• normalize unicode to combined unicode characters on field level (text)
• provide a mechanism to suppress normalization

[gogobd]

I was looking into that at the Alpine City Plone Sprint and tracked the problem down to the way unicodes are being handled by Python, Zope and Plone. First I was looking into Zope's HTTPRequest and learned that its _decode function isn't being used by Plone under all circumstances. Further I learned that class ZopeFieldStorage uses Python cgi's FieldStorage which is - and should under all circumstances stay - completely agnostic of any encodings. It's just handling binary data.

So handling incoming unicode in the request(s) would turn out to be quite invasive. And I abandoned that approach.

The proposed solution of handling unicode appropriately on field level immediately looked better. Unicode handling in zope.schema is pretty much straight forward and I was able to isolate the problem to just one class: Text, implementer of IFromUnicode. My proposed fix would also provide a flag to suppress unicode normalization for the (in my humble opinion rare) occasions where it is unwanted.

I will create a pull request, see #87.

[jamadden]

Fixed in #87 and #90