Verbatim-safe typography, ^-punctuation quoting, full-range ^UUUU^, polyglot html

Typographic transforms (---, quote pairs, ~) no longer touch verbatim
text: @c/@code/@source_listing content and ^'...'^ spans show exactly the
characters written. "^" before any punctuation character quotes it in
every target (the apostrophe excepted: ^' opens a literal span), with the
new :resolve option on @@@target declaring per-target renderings. The
^UUUU^ code-point form accepts 4-6 hex digits, the full Unicode range.
The html output and transform spellings are polyglot (XML-valid), in
preparation for an EPUB target. New suites: transform_test, character_test
(engine), typography_test (SKS).

(from dev 07ce5ea86a0a)

mac/character.h

@@ -11,7 +11,11 @@
 #include <regex>
 #include <unistd.h>
 
-const std::regex unicode_re(R"(\^(([0-9A-Fa-f]{5})|([0-9A-Fa-f]{4})|([0-9A-Fa-f]))\^)");
+// ^UUUU^ — a Unicode code point in hex, 4 to 6 digits (^263A^ is the BMP,
+// ^13000^ EGYPTIAN HIEROGLYPH A001, ^10FFFD^ the top of the range), or a
+// single digit.  Lengths 2-3 are NOT accepted: two- and three-letter
+// sequences of a-f collide with the ^s^-style mnemonic names.
+const std::regex unicode_re(R"(\^(([0-9A-Fa-f]{4,6})|([0-9A-Fa-f]))\^)");
 const std::regex unicode_hide_re(R"(=([0-9A-Fa-f]{2})=)");
 const std::regex hex2_re(R"(([0-9A-Fa-f]{2}))");
 const std::regex hex4_re(R"(([0-9A-Fa-f]{4}))");