This patch has revised the callbacks in scm/tablature.scm. It appears
to work properly.
The tab-tie-follow-engraver is still using SCM calls to analyze the
grobs. I will get that worked out soon, I hope (although I seem to have
gone backward, not forward, today).
In this patch, the glissando, slur, tie, and repeat-tie callbacks have
been modified. The slur callback doesn't even check for the tie status,
since there's no change.
The glissando callback checks to see if the
notehead has been forced to display with parentheses, but it does not
set the notehead display format. The tie and repeat-tie callbacks are
responsible for setting the notehead display.
I think that this is the right logic. Because the tie can be split, the
tie (and repeat-tie) callbacks need to decide on the note head display.
The slur and glissando callbacks don't need to set the display, they
just respond to it if need be.