optimize: only re-insert pngs after pngquant

Previously we attempted to reinsert all PNGs, but it appears to be
unlikely that Leptonica's API is actually capable of optimizing the PNG
before it inserts it.

In any event qpdf has gained image optimization capabilities as well
which we coudld borrow.

src/ocrmypdf/optimize.py

@@ -346,6 +346,7 @@ def transcode_jpegs(pike, jpegs, root, log, options):
 
 
 def transcode_pngs(pike, images, image_name_fn, root, log, options):
+    modified = set()
     if options.optimize >= 2:
         png_quality = (
             max(10, options.png_quality - 10),
@@ -366,6 +367,7 @@ def transcode_pngs(pike, images, image_name_fn, root, log, options):
                         png_quality[1],
                     )
                 )
+                modified.add(xref)
             with tqdm(
                 desc="PNGs",
                 total=len(futures),
@@ -375,7 +377,7 @@ def transcode_pngs(pike, images, image_name_fn, root, log, options):
                 for _future in concurrent.futures.as_completed(futures):
                     pbar.update()
 
-    for xref in images:
+    for xref in modified:
         im_obj = pike.get_object(xref, 0)
         try:
             compdata = leptonica.CompressedData.open(png_name(root, xref))