Enhance gabriel.view to render paraphrases and media inputs

[hemanth-asirvatham]

Motivation

• Make it easy to compare paraphrased outputs with their originals inside the notebook viewer by detecting revised columns and rendering a clear paraphrase / original split.
• Improve UX for non-text modalities by rendering images and audio inline (centered image or audio player) instead of showing raw file paths or base64 strings.
• Preserve existing behavior when no paraphrase or media is present so the current viewer UI remains unchanged.

Description

• Added media/paraphrase-related imports and helpers including encode_audio and load_image_inputs and an os import for path handling in src/gabriel/utils/passage_viewer.py.
• Introduced _resolve_paraphrase_columns to detect paraphrase/revised columns (including numbered _<n> variants) and _media_from_value to detect images/audio from URLs, data URIs, paths, or base64 blobs.
• Implemented _build_media_html, _build_paraphrase_html, and _build_passage_body to render media blocks and a two-section paraphrase view that shows Paraphrased text (emphasized) above a thin divider and Original text below.
• Wire-up: the passage payload now carries revised and media keys; the main rendering path uses _build_passage_body so existing highlighted-snippet rendering is preserved when no paraphrase/media are found; added CSS rules to style paraphrase sections and media presentation for both dark and light themes.

Testing

• No automated tests were run as part of this change.

---

Codex Task

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}