Docs: Add format v3 implementation status

[manuzhang]

Summary

• consolidate V1-V3 table and catalog operation matrices with a Spec column
• retain separate rows only when an operation's status or applicability differs by format version
• add V3-specific capability coverage and refresh current data type and operation status across all five libraries

The audit found that the old V1/V2 maintenance, data-read, and Puffin-planning status differences were stale rather than version-gated. The remaining version splits are intentional: PyIceberg's V3 metadata write gap changes table-update and append-write support, position-delete writes are V2-only, and deletion vectors are V3-only.

A Y in the V3 capability matrix requires end-to-end support for the listed behavior rather than metadata serialization alone. A range in an operation table's Spec column means the status is identical for every listed version.

This supersedes #17311 and incorporates its review feedback by removing duplicated, inconsistent rows for operations that are not format-version-specific.

Closes #17308.

Audit scope

Implementation heads checked on 2026-09-11:

• Java: 954a22296
• PyIceberg: 1652c1a0f
• Rust: 44ceed301
• Go: 218373648
• C++: c136a1f92

Testing

• make -C site lint
• git diff --check

---

AI Disclosure

• Model: GPT-5
• Platform/Tool: Codex
• Human Oversight: unreviewed
• Prompt Summary: Audit Apache Iceberg format V3 support across language implementations and update the implementation status page.