Extend REGEX_CHARS with more "reserved" regex chars.

Extend REGEX_CHARS with more "reserved" regex chars.

Added the following three characters to the list of "reserved" regular
exprssion chars: \ " $ . The problem fixes searches for some Chinese
terms that are normalized with ICU. Indeed these terms include one
of the above REGEX_CHARS. If they are not properly escaped it will
reproduce invalid regular expressions.