update prior master-dupe pairings

[taimursajid]

When adding a new match, prior matches made to master and dupe pairs will also be added to _matches_list so that new matches flow through.

The following toy example demonstrates how the issue:

sample = [
    'microsoftoffice 365 home',
    'microsoftoffice 365 pers',
    'microsoft office'
    ]

df = pd.DataFrame(sample, columns=['name'])

sg = StringGrouper(df['name'])
sg = sg.fit()

sg = sg.add_match('microsoft office','microsoftoffice 365 home')
sg = sg.add_match('microsoftoffice 365 pers','microsoft office')
df['deduped'] = sg.get_groups()

The existing code will not make add a master-dupe entry for 'microsoftoffice 365 pers' and 'microsoftoffice 365 home' in _matches_list, so microsoftoffice 365 pers will map to itself.

A small change has also been added to the _matches_list so that duplicates are dropped.

[Bergvca]

This looks great! Thanks for finding and fixing this bug. Can I ask you to add a unittest? You could use the example above for example. If not I can create one later.

[taimursajid]

Done.

[Bergvca]

Merged and updated in pypi (version 0.1.2). Thank you very much!