* signal() sets SA_RESTART by default.
* Some syscalls can theoretically block indefinitely.
  Even opening a special file could result in an indefinitely
  blocking operation, that should be interruptible.
  You must still poll teco_interrupted in these read() loops of course.
* Also makes sure that clipboard operations are interruptible
  even if $SCITECO_CLIPBOARD_GET blocks.
  Although I couldn't provoke problems in practice,
  I did observe hangs with xclip on Wayland on Linux,
  that could only be resolved by manually killing xclip.

* Hopefully also fixed FreeBSD CI builds. Always define NCURSES_WIDECHAR.
* Windows builds were probably also broken.
  We now pass down PDC_WIDE and PDC_FORCE_UTF8 into Scinterm as well.

Also, Scinterm had to be updated again since NetBSD curses stores up to 8 wchar_t per cell.

* We fork after command line arguments have been parsed, which
  is after gtk_get_option_group() has been called.
  This means that GTK was already initialized and it wasn't safe
  to continue after forking.
* As a workaround, we now re-exec with the original argv array,
  so GTK can be properly reinitialized.
  Since we did not remove `--detach` from argv (and that would be
  nontrivial), it would fork again endlessly,
  so we use an environment variable
  $__SCITECO_DETACHED to guard against recursive forks.
* Also, do not close stdin/stdout/stderr if has been redirected
  to a file, so you can now e.g. call `gsciteco -d >some-file`.
* This was broken since v2.5.0.

* Turns out that it works perfectly with the default --enable-malloc-replacement
  as malloc() is replaced globally.
* Added note to configure.ac that malloc() replacement may be useful on platforms
  without a global symbol namespace - you just have to link statically (--enable-static-executables)
  as well.
  This wasn't necessary on Haiku, though.
* Only the test suite required a minor fix since Haiku doesn't support `ulimit`.
* The Gtk interface is broken though on Haiku: You cannot type dead keys,
  Ctrl and AltGr combinations.
  Doesn't appear to be SciTECO-specific though. The libraries simply doesn't
  report modifiers. Appears to be a bug in Haiku's GTK 3 port,
  at least in their Beta5 branch. It can be reproduced with gtk-demo as well.
* This opens the door for contributing a port into HaikuPorts.
  This will probably be Curses-only for the time being (see above).

* While xcurses-config does define PDC_WIDE, it does not
  define PDC_NCMOUSE, which we currently rely on so that
  NCURSES_MOUSE_VERSION is set correctly.
  Therefore we check for it just like when using --with-interface=pdcurses.
* The modifiers were broken on all variants of PDCurses.
  This was a regression from v2.4.0.

This is useful to launch from a terminal without "blocking" this terminal.
There are tools like nohup and daemonize (BSD) to do the same, but having it
builtin is shorter to write.