Add document chunks SDK methods

[suguanYang]

Summary

• Add client.documents.list_chunks(document_id, ...) for GET /v1/documents/{document_id}/chunks.
• Add client.documents.get_chunk(document_id, document_chunk_id, ...) for GET /v1/documents/{document_id}/chunks/{document_chunk_id}.
• Add Pydantic response models, sync/async resource methods, README/docs usage, and focused resource tests.
• Keep asset URL generation opt-in via include_asset_urls only.

Verification

• uv run --python 3.11 --extra dev pytest tests/test_documents.py
• uv run --python 3.11 --extra dev mypy src/knowhere
• uv run --python 3.11 --extra dev ruff check src/knowhere tests/test_documents.py
• git diff --check
• Live staging smoke against https://api-staging.knowhereto.ai:
  • client.documents.list(namespace="default") returned 9 documents.
  • client.documents.list_chunks(document_id, page=1, page_size=2) returned 2 chunks with total 651.
  • client.documents.get_chunk(document_id, chunk_id) returned the same chunk.
  • include_asset_urls=True returned asset URLs for sampled image and table chunks.

Upstream API

• Follows merged API PR https://github.com/Ontos-AI/knowhere-api/pull/227