Check if regex boundary is needed

[tabergma]

Description of Problem:
Currently, we are adding a \b (boundary) to every regex for the RegexFeaturizer and RegexEntityExtractor. In #7091 a user mentioned that this is not working for non-whitespace tokenizable language, such as Chinese.
The question is, do we actually need to add the boundary at all? (See slack thread for reference.)

Overview of the Solution:
Figure out why the boundary was added in the first place and check if it is still needed. If not, remove it from the code.

[tabergma]

related to #7421