We found that the cleanser inside of the Match rule doesn't cleanse some characters, although their cleansing is defined in the cleanser dictionary.
In the dictionary file replacement defined this way:
Þ => th
Ð => D
The cleanser doesn't truncate spaces before and after the '=>' operator. This means in this case the cleanser will try to find 'Þ ' and replace it with ' th' (with spaces). To resolve this issue we need to do this:
1) Download the dictionary and remove spaces:
2) Upload the dictionary with a new name.
3) Configure this name in the cleanser inside of the match rule in the tenant L3 configuration.
4) Run rebuild match table task for the affected entities or the whole data set.