Encoding bug

[Bjwebb]

When CSV2IATI-backend opens a CSV file it is an encoded string, which is later decoded with the relevant character encoding. However, when grouping the data (if the multiple option is enabled) a decdoded (python unicode type) string is used as a key of undecoded data. This works provided the encoding is simply ascii, but fails otherwise. https://github.com/IATI/CSV2IATI-backend/blob/master/csviati/__init__.py#L187

Example error (where a unicode BOM is the non-ascii character):

KeyError: u'\ufeffproject_ID'

[Bjwebb]

This particular issue has been fixed on master. Remaining task:

• [ ] Check for this same issue elsewhere in the code