Make apply_stch() give a more precise fit

This aims to make Syriac Abbr Mark sizing more accurate when repeating segments are used, by adding an extra repeat and tightening up the spacing slightly rather than leaving a shortfall corresponding to a partial repeat-width.

src/hb-ot-shape-complex-arabic.cc

diff --git a/src/hb-ot-shape-complex-arabic.cc b/src/hb-ot-shape-complex-arabic.cc
index 5075477..6d9d093 100644
--- a/src/hb-ot-shape-complex-arabic.cc
+++ b/src/hb-ot-shape-complex-arabic.cc
@@ -542,7 +542,18 @@
 
       hb_position_t w_remaining = w_total - w_fixed - overlap;
       if (sign * w_remaining > sign * w_repeating && sign * w_repeating > 0)
-	n_copies = (sign * w_remaining + sign * w_repeating / 4) / (sign * w_repeating) - 1;
+	n_copies = (sign * w_remaining) / (sign * w_repeating) - 1;
+
+      /* See if we can improve the fit by adding an extra repeat and squeezing them together a bit. */
+      hb_position_t extra_repeat_overlap = 0;
+      hb_position_t shortfall = sign * w_remaining - sign * w_repeating * (n_copies + 1);
+      if (shortfall > 0)
+      {
+        ++n_copies;
+        hb_position_t excess = (n_copies + 1) * sign * w_repeating - sign * w_remaining;
+        if (excess > 0)
+          extra_repeat_overlap = excess / (n_copies * n_repeating);
+      }
 
       if (step == MEASURE)
       {
@@ -568,6 +579,8 @@
 	  for (unsigned int n = 0; n < repeat; n++)
 	  {
 	    x_offset -= extents.width;
+	    if (n > 0)
+	      x_offset += extra_repeat_overlap;
 	    pos[k - 1].x_offset = x_offset;
 	    /* Append copy. */
 	    --j;