Add TODOs for getTokens() to known places that scan command lines.

Lacking time to properly test new LLStringUtil::getTokens() against the
present (different!) command-line scanners in LLExternalEditor::tokenize() and
LLCommandLineParser::parseCommandLineString(), just annotate as future work
the goal of unifying them... SIGH.

indra/newview/llcommandlineparser.cpp

@@ -342,6 +342,15 @@ bool LLCommandLineParser::parseCommandLine(int argc, char **argv)
     return parseAndStoreResults(clp);
 }
 
+// TODO:
+// - Break out this funky parsing logic into separate method
+// - Unit-test it with tests like LLStringUtil::getTokens() (the command-line
+//   overload that supports quoted tokens)
+// - Unless this logic offers significant semantic benefits, replace it with
+//   LLStringUtil::getTokens(). This would fix a known bug: you cannot --set a
+//   string-valued variable to the empty string, because empty strings are
+//   eliminated below.
+
 bool LLCommandLineParser::parseCommandLineString(const std::string& str)
 {
     // Split the string content into tokens

indra/newview/llexternaleditor.cpp

@@ -119,6 +119,12 @@ std::string LLExternalEditor::getErrorMessage(EErrorCode code)
 	return LLTrans::getString("Unknown");
 }
 
+// TODO:
+// - Unit-test this with tests like LLStringUtil::getTokens() (the
+//   command-line overload that supports quoted tokens)
+// - Unless there are significant semantic differences, eliminate this method
+//   and use LLStringUtil::getTokens() instead.
+
 // static
 size_t LLExternalEditor::tokenize(string_vec_t& tokens, const std::string& str)
 {