[GPOS] Sanitize Cursive positioning anchors lazily

Speeds up Duployan-Regular.otf load time by 30%.
Doesn't seem to slow down shaping in a measurable way.

src/OT/Layout/GPOS/CursivePosFormat1.hh

@@ -91,7 +91,13 @@ struct CursivePosFormat1
   bool sanitize (hb_sanitize_context_t *c) const
   {
     TRACE_SANITIZE (this);
-    return_trace (coverage.sanitize (c, this) && entryExitRecord.sanitize (c, this));
+    if (unlikely (!coverage.sanitize (c, this)))
+      return_trace (false);
+
+    if (c->lazy_some_gpos)
+      return_trace (entryExitRecord.sanitize_shallow (c));
+    else
+      return_trace (entryExitRecord.sanitize (c, this));
   }
 
   bool intersects (const hb_set_t *glyphs) const
@@ -120,6 +126,7 @@ struct CursivePosFormat1
 
     const EntryExitRecord &this_record = entryExitRecord[(this+coverage).get_coverage  (buffer->cur().codepoint)];
     if (!this_record.entryAnchor) return_trace (false);
+    if (unlikely (!this_record.sanitize (&c->sanitizer, this))) return_trace (false);
 
     hb_ot_apply_context_t::skipping_iterator_t &skippy_iter = c->iter_input;
     skippy_iter.reset_fast (buffer->idx, 1);
@@ -136,6 +143,7 @@ struct CursivePosFormat1
       buffer->unsafe_to_concat_from_outbuffer (skippy_iter.idx, buffer->idx + 1);
       return_trace (false);
     }
+    if (unlikely (!prev_record.sanitize (&c->sanitizer, this))) return_trace (false);
 
     unsigned int i = skippy_iter.idx;
     unsigned int j = buffer->idx;