Add User filter on wiki page

[SaulLu]

This PR add a filter that filters out all the examples whose "title" field insed their "meta" value starts with "User ". The goal here is to remove these type of examples:

"Users in this category indicate they have knowledge of language Nepali."

"Users in this category indicate they have skill level 1 for language Low Saxon."

I've tested it on lm_en_wikinews_filtered, here's the logs:

03/03/2022 12:33:09 - INFO - __main__ - Applied filter: filter_user_titles
03/03/2022 12:33:09 - INFO - __main__ -      Initial number of samples: 54387 samples
03/03/2022 12:33:09 - INFO - __main__ -      Removed samples: 345 samples
03/03/2022 12:33:09 - INFO - __main__ -      Removed percentage: 0.63 %

I've also checked by hand all the filtered out examples and all of them corresponds to examples that we want to filter out.

Partially solves #5

Remaining todo: have the list of datasets we want this filter apply on

[lvwerra]

LGTM!