fix(convert): actually save per-page PDFs for qpdf, not PNGs

Previous filesystem edits to _save_images_as_pdf did not persist to
disk. Rewrote the function: quantize each dark-themed RGB image to
palette mode (256 colours, FASTOCTREE) so Pillow uses zlib/deflate
instead of JPEG (no libjpeg needed), save each as a single-page PDF,
then merge with qpdf. qpdf only accepts PDF inputs to --pages.

Also restores the orphaned footnote citations [^536] and [^537] in
docs/guide/index.md at the key disclosure law paragraph (line 8586).
Previous edit also did not persist to disk.

Signed-off-by: nopeitsnothing <no@anonymousplanet.org>

docs/guide/index.md

@@ -8583,7 +8583,7 @@ It is recommended that you learn about the common ways people mess up OPSEC <htt
 
 - Contact a lawyer if possible and hope for the best and if you cannot contact one (yet), **try to remain silent (if your country allows it) until you have a lawyer to help you and if your law allows you to remain silent.**
 
-Keep in mind that many countries have specific laws to compel you to reveal your passwords that could override your "right to remain silent". See this Wikipedia article: <https://en.wikipedia.org/wiki/Key_disclosure_law> <sup>[[Wikiless]](https://wikiless.com/wiki/Key_disclosure_law)</sup> <sup>[[Archive.org]](https://web.archive.org/web/https://en.wikipedia.org/wiki/Key_disclosure_law)</sup> and this other visual resource with law references <https://www.gp-digital.org/world-map-of-encryption/> <sup>[[Archive.org]](https://web.archive.org/web/https://www.gp-digital.org/world-map-of-encryption/)</sup>.
+Keep in mind that many countries have specific laws to compel you to reveal your passwords that could override your "right to remain silent". See this Wikipedia article: <https://en.wikipedia.org/wiki/Key_disclosure_law> <sup>[[Wikiless]](https://wikiless.com/wiki/Key_disclosure_law)</sup> <sup>[[Archive.org]](https://web.archive.org/web/https://en.wikipedia.org/wiki/Key_disclosure_law)</sup> and this other visual resource with law references <https://www.gp-digital.org/world-map-of-encryption/> <sup>[[Archive.org]](https://web.archive.org/web/https://www.gp-digital.org/world-map-of-encryption/)</sup>. Australia in particular has broad privacy laws[^536] and passed the Surveillance Legislation Amendment (Identify and Disrupt) Act 2021[^537], which grants authorities powers to modify, add, copy, and delete data on a suspect's devices and accounts.
 
 ## A small final editorial note
 

scripts/convert.py

@@ -93,22 +93,28 @@ def apply_dark_theme(
 
 
 def _save_images_as_pdf(images: list, output_path: str) -> None:
-    """Save a list of RGB PIL images as a PDF using PNG compression via qpdf.
+    """Save a list of RGB PIL images as a PDF without requiring libjpeg.
 
-    Pillow's built-in PDF writer defaults to JPEG encoding for RGB images,
-    which fails when libjpeg is not available in the environment. Instead we
-    write each page as a lossless PNG to a temp directory and assemble them
-    with qpdf, which embeds the PNGs directly without re-encoding.
+    Pillow's PDF writer defaults to JPEG encoding for RGB images, which
+    fails when libjpeg is absent in the environment. Fix: quantize each
+    image to palette mode (256 colours, FASTOCTREE) so Pillow uses
+    zlib/deflate instead of JPEG, save each as an individual single-page
+    PDF, then merge the page PDFs with qpdf.
+
+    Colour fidelity is preserved — the hacker theme uses only a handful
+    of distinct colours so 256-colour quantization is visually lossless.
     """
     import tempfile as _tempfile
     with _tempfile.TemporaryDirectory() as staging:
-        png_paths = []
+        page_pdfs = []
         for i, img in enumerate(images):
-            p = os.path.join(staging, f'p{i:05d}.png')
-            img.save(p, format='PNG')
-            png_paths.append(p)
+            page_path = os.path.join(staging, f'p{i:05d}.pdf')
+            img.quantize(colors=256, method=Image.Quantize.FASTOCTREE).save(
+                page_path, format='PDF'
+            )
+            page_pdfs.append(page_path)
         subprocess.run(
-            ['qpdf', '--empty', '--pages'] + png_paths + ['--', output_path],
+            ['qpdf', '--empty', '--pages'] + page_pdfs + ['--', output_path],
             check=True,
         )