Return empty text handle on no match for RegexTextExtractor

[bglearning]

Is your feature request related to a problem? Please describe.
If RegexTextExtractor doesn't find a match, it returns an empty dict {} which cannot be processed further in the pipeline (?), essentially dead-ends.

Describe the solution you'd like
Would be good to return {"captured_text": ""} instead which can then be processed downstream (e.g. through a ConditionalRouter that checks captured_text). Alternatively, an explicit no_match output could also be fine.

The run method docstring mentions return_empty_on_no_match (which when False would return the empty match) but that is not a param for the class.