* The previous convention of !* ... *! are now true block comments,
  i.e. they are parsed faster, don't spam the goto table and allow
  embedding of exclamation marks - only "*!" terminates the comment.
* It is therefore now forbidden to have goto labels beginning with "*".
* Also support "!!" to introduce EOL comments (like C++'s //).
  This disallows empty labels, but they weren't useful anyway.
  This is the shortest way to begin a comment.
* All comment labels have been converted to true comments, to ensure
  that syntax highlighting works correctly.
  EOL comments are used for single line commented-out code, since it's
  easiest to uncomment - you don't have to jump to the line end.
  This is a pure convention / coding style.
  Other people might do it differently.
* It's of course still possible to abuse goto labels as comments
  as TECO did for ages.
* In lexing / syntax highlighting, labels and comments are highlighted differently.
* When syntax highlighting, a single "!" will first be highlighted as a label
  since it's not yet unambiguous. Once you type the second character (* or !),
  the first character is retroactively styled as a comment as well.

* g_utf8_get_char_validated() returns -2 for null-bytes (sometimes!?)

* Apparently g_utf8_get_char_validated() sometimes(!) returns -2 for null-characters,
  so it was considered an invalid byte sequence.
* What's strange and unexplainable is that other uses of the function, as are behind nA and nQq,
  did not cause problems and returned 0 for null-bytes.
* This also fixes syntax higlighting of .teco_session files which use the null-byte as the
  string terminator.
  (.teco_session files are not highlighted automatically, though.)

This is a purely cosmetic change.

* The teco_qreg_vtable_t::get_string() method should support returning the
  length optionally (may be NULL).
  This already worked with teco_doc_get_string(), even though it wasn't documented,
  and therefore didn't cause problems with regular Q-Registers.

* @ES/SCI_SETILEXER/lib^@name/ now opens the lexer <name> in library <lib>.
* You need to define the environment variable $SCITECO_SCINTILLUA_LEXERS to point
  to the lexers/ subdirectory (containing the *.lua files).
  Perhaps this should default to the dirname of <lib>?
* The semantics of SCI_NAMEOFSTYLE have been changed:
  It now returns style ids when given style names, so you can actually write
  Scintillua lexer *.tes files.
  This will be superfluous if we had a way to return strings from Scintilla messages into
  Q-Registers, e.g. 23@EPq/SCI_NAMEOFSTYLE/.
* We now depend on gmodule as well, but it should always be part of glib.
  It does not change the library dependencies of any package.
  It might result in gmodule shared libraries to be bundled in the Win32 and Mac OS
  packages if they weren't already.

* checks for character consistency (of UTF-8 byte sequences) were slowing down things significantly in Scintilla
* It got even worse if the file indeed contained non-ANSI codepoints as reading in chunks of 1024
  would sometimes mean that incomplete byte sequences would be read.
  Some large 160mb test files wouldn't load even after minutes.
  They now load in seconds.
* This does NOT yet solve the slowdowns when operating on very long lines.

* I.e. you can now write FK...$^YD to delete up to AND the matched pattern.

* SCI_SETILEXER(NULL) is not a reliable way to do that since
  that's the default for all views.
* This was breaking the git.tes lexer for instance and was unnecessarily
  driving teco_lexer_style() on plain-text documents.
* Since we currently do not implement the ILexer5 C++ interface
  and teco_view_t is just a pointer alias, we are abusing the view's "identifier" instead.
  This is probably sufficient, as long as there is only one lexer "in the container".
  Otherwise, there should perhaps be a single C++ class that does nothing but
  wrapping a callback into an ILexer5 object with a C ABI.