docs: add dev-notes post for the Nemotron Indic translation plugin (#959)

* docs: add dev-notes post for the Nemotron Indic translation plugin

Adds a draft dev-notes post covering data-designer-nemotron-indic-translation:
the translation pipeline's three paths (direct LLM translation, windowed
translation for Santali, script-remap for Manipuri), model routing, and
romanization/native-numeral handling. Includes Graphviz-generated figures
for each path and registers new author entries.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: register the gemma model alias in the Try It Yourself example

DataDesignerConfigBuilder() was constructed with no model_configs, so
model_alias="gemma"/evaluator_model_alias="gemma" had no alias to
resolve and preview() would fail at model lookup. Registers "gemma" via
ModelConfig with the default "nvidia" provider, matching the pattern
used in docs/notebook_source/1-the-basics.py.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: add landing page card for the Nemotron Bharat dev-notes post

The nav entry in latest.yml made the post reachable, but the separately
maintained Dev Notes landing grid (devnotes/index.mdx) never got a
BlogCard for it, so readers browsing the overview wouldn't see it
alongside the other posts.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: use a generic threshold label in the Path 1 diagram

The diagram hardcoded "score >= 75?", implying one threshold for all
20 languages on this path. The article states the 8 lower-resource
languages on this path use a lower, per-language threshold, so the
hardcoded 75 misrepresented when those translations pass evaluation.
Changed to a generic "score >= threshold?" label, matching the alt
text and the plugin's rewrite_threshold field.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: drop the print() wrapper around display_sample_record()

display_sample_record() returns None and renders via rich as a side
effect, so print(...) around it emits a stray "None" after the
rendered record. All other call sites across the docs invoke it bare.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: add Yev Meyer as an author on the Nemotron Bharat post

Added last in the byline and the landing-page card, using his existing
.authors.yml title (Principal Research Scientist at NVIDIA), matching
how he's credited on his other dev-notes posts.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: add hero image to the Nemotron Bharat dev-notes post

Adds an abstract illustration of the three translation paths (direct,
windowed, script-remap) fanning from a source document into translated
outputs. Wired into both the post itself (right after the excerpt
marker) and the landing-page BlogCard, matching how other dev-notes
posts use hero images. The imagery avoids implying any particular
script or language.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>

* docs: address review comments on the Nemotron Bharat dev-notes post

Clarify the no-newline fallback for windowing, list the Meitei Mayek
generation issues, and explain why native-numeral prose is used.

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

* docs: tighten the windowing paragraph in the Nemotron Bharat post

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

* docs: note that the gemma and bodhan model aliases are illustrative

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

* docs: shorten the Nemotron Bharat nav title so the URL slug is reasonable

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

* docs: import DataDesignerConfigBuilder from the top-level config namespace

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

* docs: align rubric names and fix the config-walkthrough transition

Use the plugin's actual rubric dimension names in the intro, and reword
the transition paragraph so it covers romanization and native numerals.

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

* docs: add config examples for romanization and native numerals

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

* docs: use numeric-ID avatar URLs for the Nemotron Bharat authors

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

* docs: update Niket Agarwal's title to Senior Distinguished Engineer

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

* docs: update author titles for the Nemotron Bharat post

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

* docs: make the Try It Yourself example runnable and clarify numeral style

Use the lowercase NVIDIA API model ID, replace the seed placeholder with a
LocalFileSeedSource path, and describe the numeral-style column as the
requested style rather than a verified one.

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

* docs: rename Nemotron Bharat to Nemotron Indic in display text and assets

Rename the display names, page slug, and assets directory. The plugin
package name, GitHub URLs, and the post file name are unchanged.

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

* docs: use the renamed nemotron-indic-translation plugin package

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

---------

Signed-off-by: Kiran Praveen <kipraveen@nvidia.com>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

Author: kipraveen