Remove dependency on pandas

[timcu]

Currently pandas is used purely to read the excel file and pandas uses xlrd to do all the work anyway. Installing pandas on some computers is difficult (eg raspberry pi running raspbian) so if we could remove the requirement for pandas then the software could be deployed in more places requiring less resources. Documentation for xlrd says it is best for .xls files but openpyxl is best for .xlsx files. Unfortunately source data file can currently be either format. May need to use both libraries until gov stops using .xls files.

[anniequasar]

so remove pandas and replace it with xlrd and openpyxl for now?

[timcu]

Should be able to use just xlrd. I don't think openpyxl can do xls files but xlrd can do both xls and xlsx files.

[timcu]

I have replaced pandas with xlrd in the tests

[timcu]

To replace pandas in datasheet.py you will need to load workbook from a http response rather than from a file. xlrd doesn't do that out of the box but you can combine it with urllib to do the following

book = xlrd.open_workbook(file_contents=urlopen(find_xl_url()).read())