fix: keep redaction marks correct with custom masks, Prism and Mermaid

- The mask from /api/options is stored once loaded; code markers and
  rendered Markdown are refreshed when it arrives instead of keeping the
  XXXX default.
- Mermaid sources are left untouched so diagrams still parse.
- Fenced code blocks are marked from a Prism complete hook, after Prism
  has rewritten their HTML.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

scripts/dev-mock.js

@@ -292,7 +292,7 @@ const explorerFiles = {
   "src/utils": [{ name: "data.py", path: "src/utils", size: 300, sha: "c1" }],
 };
 const explorerContent = {
-  "README.md": "# DeepLearnUtils\n\nDeveloped by XXXX-1 at XXXX-2.\n\nUtilities for reproducing the paper's experiments. Every script in `src/` reads its configuration from `configs/` and writes results to `out/`, so a full run of the evaluation only needs the two commands below. The long sentence here checks the reading width of rendered Markdown.\n\n## Quick start\n\n```bash\npip install -r requirements.txt\npython src/train.py --config configs/base.yaml\n```\n\n| Model | Accuracy |\n| --- | --- |\n| Baseline | 71.2 |\n| Ours | 78.9 |\n\nContact: XXXX-3\n",
+  "README.md": "# DeepLearnUtils\n\nDeveloped by XXXX-1 at XXXX-2.\n\nUtilities for reproducing the paper's experiments. Every script in `src/` reads its configuration from `configs/` and writes results to `out/`, so a full run of the evaluation only needs the two commands below. The long sentence here checks the reading width of rendered Markdown.\n\n## Quick start\n\n```bash\n# maintained by XXXX-1\npip install -r requirements.txt\npython src/train.py --config configs/base.yaml\n```\n\n```mermaid\ngraph LR; XXXX-2 --> Paper\n```\n\n| Model | Accuracy |\n| --- | --- |\n| Baseline | 71.2 |\n| Ours | 78.9 |\n\nContact: XXXX-3\n",
   LICENSE: "MIT License\n\nCopyright (c) 2026 XXXX-1\n",
   "requirements.txt": "torch>=2.3\nnumpy\npyyaml\n",
   "src/train.py": '# Copyright 2026 XXXX-1, XXXX-2\nimport argparse\nimport yaml\n\nfrom utils.data import load_dataset\n\n\ndef main():\n    parser = argparse.ArgumentParser(description="Train the model")\n    parser.add_argument("--config", required=True)\n    args = parser.parse_args()\n    with open(args.config) as f:\n        config = yaml.safe_load(f)\n    dataset = load_dataset(config["data"])\n    print(f"Loaded {len(dataset)} examples")\n\n\nif __name__ == "__main__":\n    main()\n',