While here, also drop those calls to memset, as the whole info object is
being zeroed at the beginning of main(). (Probably this is useful upon
config reload, too.)
Also add a little hack to $scroll to omit the need to call new_fg() from
inside print_scroll(). Instead of inserting a new special after printing
the scrolled text, insert a new object after the scroll object when
parsing text objects to handle the color reset.
The callback functions are:
print - to be called from generate_text_internal()
iftest - same as print for ifblock objects (triggers jumping)
barval - same as above for bar objects (returns current bar value)
gaugeval - same as above for gauge objects
graphval - same as above for graph objects
percentage - for percentage objects, returns actual percentage
free - called in free_text_objects()
Until conversion is complete, if the function pointer is NULL the old
lookup by type is being done.
Note that it's possible to assign both 'print' and 'iftest' callbacks.
In this case, the code simply ignores the 'iftest' callback, though this
could easily be changed (always calling DO_JUMP at last, of course) in
order to allow ifblock objects to print something in addition to jumping
somewhere (or not).
The decision about whether to print ASCII or X11 bar is done from within
specials.c, so all those #ifdef + if () blocks can be dropped. This also
implicitly enables the ASCII bar for some bar printing objects which where
forgotten before.
Murphy hit me again: in my naive attempt to fix the clash between
ifblocks and objects parsing text objects due to the double use of the
'sub' field, I overlooked this problem with reusing the 'special_data'
field. So here comes the real thing (TM), donating ifblocks their own
field for pointing to the jump target.
