* SIGTERM used to insert the ^KCLOSE key macro.
  However with the default ^KCLOSE macro, which inserts `EX`,
  this may fail to terminate the editor if buffers are modified.
  If the process is consequently killed by a non-ignorable signal,
  we may still loose data.
* SIGTERM is used to gracefully shut down, so we now always terminate.
  Since we have recovery files, they are now dumped before terminating.
  This makes sure that recovery files are more up-to-date during
  unexpected but gracefull terminations.
* The same functionality is planned on Curses, but requires more fundamental
  changes (TODO).

5 minutes was probably a bit too conservative.

* When passing a struct that should not be modified, I usually use a const pointer.
* Strings however are small 2-word objects and they are often now already passed via separate
  `gchar*` and gsize parameters. So it is consistent to pass teco_string_t by value as well.
  A teco_string_t will usually fit into registers just like a pointer.
* It's now obvious which function just _uses_ and which function _modifies_ a string.
  There is also no chance to pass a NULL pointer to those functions.

* Requiring state callbacks by generating their names (e.g. NAME##_input) has several disadvantages:
  * The callback is not explicitly referenced when the state is defined.
    So an unintroduced reader will see some static function, which is nowhere referenced and still
    doesn't cause "unused" warnings.
  * You cannot choose the name of function that implements the callback freely.
  * In "substates" you need to generate a callback function if you want to provide a default.
    You also need to provide dummy wrapper functions whenever you want to reuse some existing
    function as the implementation.
* Instead, we are now using static assertions to check whether certain callbacks have been
  implemented.
  Unfortunately, this does not work on all compilers. In particular GCC won't consider
  references to state objects fully constant (even though they are) and does not allow
  them in _Static_assert (G_STATIC_ASSERT). This could only be made to work in newer GCC
  with -std=c2x or -std=gnu23 in combination with constexpr.
  It does work on Clang, though.
  So I introduced TECO_ASSERT_SAFE() which also passes if the expression is *not* constant.
  These static assertions are not crucial - they do not check anything that can differ between
  systems. So we can always rely on the checks performed by FreeBSD CI for instance.
  Also, you will of course quickly notice missing callbacks at runtime - with and without
  additional runtime assertions.
* All mandatory callbacks must still be explicitly initialized in the TECO_DEFINE_STATE calls.
* After getting rid of generated callback implementations, the TECO_DEFINE_STATE macros
  can finally be qualified with `static`.
* The TECO_DECLARE_STATE() macro has been removed. It no longer abstracts anything
  and cannot be used to declare static teco_state_t anyway.
  Also TECO_DEFINE_UNDO_CALL() also doesn't have a DECLARE counterpart.

This made problems in teco_state_replace_default_insert, where we had to
override the done_cb.

Perhaps we should avoid all generated callback names (ie. mandatory
callback implementations)?

* Other editors call "backup files" previous copies of saved files.
  This role would be served by savepoint files in SciTECO.
* Likewise filename~ would point to such a backup file.
  It therefore makes sense that savepoint files also end in tildes (.teco-n-filename~).
* Security copies of modified buffers would be called "auto-saves" (Emacs) or
  "swap files" (Vim).
  Both of these terms is IMHO misleading, so SciTECO now uses the
  term "recovery file".
* "Recovery files" are now named #filename# just like in Emacs.

* The backup mechanism is supposed to guard against crashes of SciTECO and
  unexpected program terminations (e.g. power cycling, etc.)
* In a given interval (no matter whether busy or idlying on the prompt)
  SciTECO saves all modified buffers with the filename~ (like most other editors).
  As an optimization files are not backed up if they have been backed up
  previously to avoid pointless and possibly slow file system writes.
* While the backup mechanism exists outside of the usual undo-paradigm -
  backup file creating is not bound to character input and it makes no sense
  to restore the exact state of backup files - there are some interesting
  interactions:
  * When a buffer is dirtyfied or saved that was previously backed up, it must always
    be reset to the DIRTY state on rubout, so backups are eventually recreated.
  * When a buffer is dirtyfied first (was clean), the backup file must be
    removed on rubout as well - we don't expect backup files for clean buffers.
* There is currently no automatic way to restore backup files.
  This could potentially be done by opener.tes and session.tes in the future,
  although you couldn't currently always get meaningful user feedback
  (whether he wants to restore the file).
  Perhaps we should at least log a message when detecting backup files that
  are newer than the file that is being opened.