[cff] Allocate stack inline instead of using hb_vector_t Speeds up glyph_extents and glyph_shape benchmarks for CFF by 10 to 16 percent!
2022-05-09T17:49:54