bug#81636: Fix switch-to-buffer-other-*

 > It seems that the option 'switch-to-buffer-obey-display-actions'
 > has no relevance here?  Despite the documentation speaking about
 > "switch-to-buffer commands":
 >
 >    When `switch-to-buffer-obey-display-actions' is non-nil,
 >    `switch-to-buffer' commands are also supported.
 >
 > I don't remember, but it seems these "switch-to-buffer commands"
 > are commands that call 'switch-to-buffer', but not the commands
 > that share the same name prefix 'switch-to-buffer-...'?
 > Because 'switch-to-buffer-other-...' commands don't call
 > 'switch-to-buffer'.

Now supporting can mean different things - we "support" alist members
even in presence of an overriding action.  But the real question here is
whether 'switch-to-buffer-other-window' is

- used by a package that for historical reasons provides it together
   with 'switch-to-buffer' and 'switch-to-buffer-other-frame' just to
   give users the possibility to choose among them, or

- intentionally used by a package leaving the user no choice just because
   the authors think that another window better suits the purpose of the
   call and in a sense they hardcoded that behavior.

I think most requests for changing the current behavior are for the
second case and we already changed many of them.  How to handle
instances of the first case is not entirely clear to me.  IIUC we have
no strict policy there.

martin