[Improvement] Provide Chinese (zh-CN) translation for Gravitino documentation

### What would you like to be improved?

Gravitino currently provides documentation only in English. There is a large Chinese-speaking user base for data lake/metadata tools in the Apache ecosystem, but non-English-reading users face a barrier to adoption. Providing a localized (zh-CN) documentation set would significantly lower the entry barrier for Chinese-speaking users.

### How should we improve?

I have already completed a full translation of the Gravitino official docs (127 pages) and would like to contribute them to the project.

Proposal:

Scope: Translate the existing docs/*.md content into Chinese, maintaining the same structure, image references, and code blocks (code stays in English).
Directory layout: Mirror the English docs under a docs/zh/ (or docs/i18n/zh-CN/) subtree, consistent with how other Apache projects (e.g., Dubbo, ShenYu, SkyWalking) organize i18n docs.
Maintenance: Keep translations in sync via the docusaurus i18n workflow (already used by the Gravitino website) — each English page has a corresponding zh-CN version.

Motivation:

Gravitino's docs site is built with Docusaurus, which natively supports i18n via locale-prefixed routes.
Chinese is the 2nd largest language community in the Apache big-data ecosystem.
Other Apache projects (DolphinScheduler, ShenYu, SkyWalking) already ship zh-CN docs.
Additional context

I have a complete working translation (127 pages) ready to contribute. If the community is interested, I can submit it incrementally (per doc section) to keep PRs reviewable. If there's no current plan for i18n, I'll hold and revisit later.

Start by reviewing the existing docs/*.md tree and the Gravitino website's Docusaurus i18n configuration. Compare the proposed docs/zh/ or docs/i18n/zh-CN/ layout with the current site structure and determine how locale-prefixed routes are maintained. Done means the 127 translated pages are integrated with matching structure, image references, and English code blocks, with a maintainable sync workflow.

Written by the indexing model from the issue text.