Use term "special character" instead of "symbol" or "punctuation"

Thanks for seconding the switch to "special characters" and for the resources. I'll try to write a patch for 7.x-2.x and check our list of special characters versus the OWASP one.

Do you think it would be a good or bad idea to change it in 7.x-1.x? 7.x-1.x uses "punctuation" which is the worst term of all per the paper. (They found people tend to have a very limited idea of what counts as "punctuation".) The code uses ctype_punct() for the check, which returns TRUE for non-alphanumeric non-whitespace printable ASCII characters. I think I remember though it is discouraged to change strings in a stable module because it breaks translations? I think it may be worth the change anyway.

Maybe the code should use "special_char_count" instead of "special_count". At the time "special_count" seemed consistent with "alpha_count" and "int_count", and perhaps avoids confusion with "char_count". Maybe it doesn't matter.

Patch for 7.x-1.x.

I was hoping to make a new 7.x-1.x release this week, like today: #2869115: Release 7.x-1.13. There have been a lot of bug fixes + Features integration since the last release. Not sure whether it's worth blocking on this issue.

Now with bytes

8.x-3.x is also using "punctuation". I can provide more info on the paper's findings re "special characters" vs. "punctuation" if desired.

Re-uploading because I can't get the testbot to stop trying to test on D7.

Re "punctuation":

The study reported in the NIST paper had 60 age/education-varied participants from the Washington DC area. The paper says they tended to view "punctuation" as a character type distinct from "special characters". The characters .,!;:?'" were each considered as "punctuation" by >80% of participants, but all other characters were considered as punctuation by <40% of participants. By contrast, the characters #$^@%&*~({}`)-+/<>!-=[}|\ were considered "special characters" by >80% of participants and "",;:.' were considered "special characters" by >70% of participants--a much larger set than with punctuation.

The paper doesn't make a recommendation of which term to use. "Special characters" seems much better than "punctuation" to me based on the study results above. The paper's explicit recommendations include giving a full list of characters allowed rather than partial lists. They also recommend against the jargon term "non-alphanumeric".

Went ahead and committed on 7.x-2.x. Will consider committing on 7.x-1.x in the future.

Missed a file in commit c7b511a.