I've had to work a bit at these files
The fixes I have made are:

Table7-6.csv  -  removed line breaks inside description text, as these are very hard to parse correctly

Table7-10.csv - removed blank line at the end  (''',,,,,,''')

Table8-5.csv - removed 3 blank lines at the end  (''',,,,,,''')

Table9-2.csv - removed 2 blank lines at the end  (''',,,,,,''')

Table11-2.csv - removed 5 blank lines at the end  (''',,,,,,''')

I have then added parsing tricks to deal with funny characters and unallowed labels, i.e.

            line = line.replace('\xa0', '')
            line = line.replace('\222', '')
            line = line.replace('\x93', '')
            line = line.replace('\x94', '')
            line = line.replace('\x96', '')
            line = line.replace('<sup>', '')
            line = line.replace('</sup>', '')
            line = line.replace('?', '')
            line = line.replace('""', '')
            line = line.replace('&ldquo', '')
            line = line.replace('&rdquo', '')
            line = line.replace('&deg', ' deg')
            line = line.replace('&', '')

            anid = anid.replace(' ', '')

I may have removed some useful text/markup in doing this.  The registry supports unicode so there are numerous options for reintroducing important aspects as they are required, but they will need to be found

how many of these replace string do you recognise?

I am still getting some tables failing to upload, I am using the test register instance at the moment
this is where I am today:
http://test.wmocodes.info/wmdr

I will get back to this tomorrow and update you, hopefully with a working draft on test that we can share with Joerg and Dom for review; i can move us onto a github review ticket at this stage, if that's all right with you

once we are happy, I can upload to the main site

to be continued

mark