doc(ai): sync vermeer-python-client with main

[bitflicker64]

Adds the missing vermeer-python-client quick start page (en and cn). The module is referenced by name in content/*/docs/quickstart/hugegraph-ai/_index.md:16 and content/*/docs/introduction/_index.md:89, but had no page anywhere in content/.

The page sits in the hugegraph-ai section at weight: 6, after rest-api.md. Section navigation is Hugo weight-driven, so no _index.md or SUMMARY.md change is needed. Layout and frontmatter follow content/*/docs/quickstart/client/hugegraph-client-python.md.

The module is not a stub: it has a working client, two registered API groups, five endpoint wrappers, full response models and a runnable demo. It ships no test suite, so every example was verified against src/pyvermeer/** and the bundled demo.

Page	Wrong	Changed to	Source
new page, intro	missing	The Python SDK for the Vermeer memory-first graph computing engine; the import package is pyvermeer; no Vermeer server version is pinned	vermeer-python-client/README.md:3, vermeer-python-client/pyproject.toml:50
new page, Requirements	missing	Python 3.9+ for the module alone, 3.10+ for the repo; Vermeer master over HTTP (the demo uses port 8688); deps requests, urllib3, python-dateutil, decorator, rich, setuptools	vermeer-python-client/pyproject.toml:27,29-39, pyproject.toml:27, .../demo/task_demo.py:26
new page, Installation	missing	Distribution name vermeer-python-client, version managed independently, not yet on PyPI; uv sync --extra vermeer from the repo root; editable path dependency, not a uv workspace member; standalone uv sync inside the module dir	vermeer-python-client/pyproject.toml:19-20, vermeer-python-client/README.md:7-15,22-27, pyproject.toml:41,91,93-97
new page, Connect to a Vermeer Master	missing	PyVermeerClient(ip, port, token, timeout=None, log_level="INFO") parameter table, plus the behaviour notes: token cannot be None and is sent as Authorization; omitting timeout stores None, because the client always forwards its own argument over the (0.5, 15.0) default in VermeerConfig; the base URL is always plain http://{ip}:{port}/; params is JSON-serialized into the body for every method including GET; 3 retries with backoff 0.1 on 500, 502 and 504; the console handler is fixed at INFO, so DEBUG is not printed	.../client/client.py:30-48, .../utils/vermeer_config.py:29, .../utils/vermeer_requests.py:36-38,48-52,61-70,82,96, .../utils/log.py:37,47
new page, End-to-End Example	missing	A runnable load task example based on the bundled demo, extended with task read-back and graph inspection; the password is read from the environment rather than hardcoded	.../demo/task_demo.py:22-45
new page, API Surface	missing	client.graph: get_graphs() to GET /graphs, get_graph(name) to GET /graphs/{name}. client.tasks: get_tasks() to GET /tasks, get_task(id) to GET /task/{id}, create_task(req) to POST /tasks/create. client.send_request() as the shared escape hatch. Master and worker API groups are unimplemented placeholders, even though MasterResponse and WorkersResponse exist	.../api/graph.py:26-37, .../api/task.py:25-38, .../client/client.py:47,56-57, .../api/master.py:1-16, .../api/worker.py:1-16, .../structure/master_data.py:71-82, .../structure/worker_data.py:99-111
new page, Requests and Responses	missing	TaskCreateRequest(task_type, graph_name, params) serializes graph_name as graph; all responses expose errcode / message / to_dict() with 0 ok, 1 error, -1 absent; VermeerGraph and TaskInfo field lists; timestamps parsed by python-dateutil, empty string becomes None	.../structure/task_data.py:47-142,145-156, .../structure/base_data.py:18-50, .../structure/graph_data.py:95-209, .../utils/vermeer_datetime.py:23-28
new page, Task Parameters	missing	The client does not validate params; the keys come from the engine; load then compute sequence; points at the Vermeer quick start for the parameter lists	.../api/task.py:35-38, .../structure/task_data.py:154-156
new page, Errors	missing	ConnectError, TimeOutError, JsonDecodeError, UnknownError with their trigger conditions; the HTTP status code is never checked, so callers must read errcode	.../utils/exception.py:19-44, .../utils/vermeer_requests.py:91-107
new page, Development Checks	missing	./style/code_format_and_analysis.sh from the repo root; source under vermeer-python-client/src/pyvermeer/; the module ships no test suite	vermeer-python-client/README.md:80

Follow-up left out on purpose

content/*/docs/quickstart/hugegraph-ai/_index.md has a "Next Steps" list that links every sibling page in the section. Adding vermeer-python-client.md there would make the section self-consistent, but that file is owned by the hugegraph-llm sync PR, and adding the link from either side creates a dead link until both PRs merge. Worth a small follow-up once both are in.

[imbajin]

Blocking: no. Summary: The new guide has two runnable-example issues that can mislead users during installation and task execution. Evidence: verified against the exact-head docs and the HugeGraph-AI client source.

[imbajin]

⚠️ Important: This is a single task read, not polling. create_task() can return before the load finishes, but the example immediately calls get_graph() on the next lines, so it may inspect an unloaded or failed graph while claiming an end-to-end flow. Please loop until the task reaches a terminal success/failure state, handle failure, and mirror the fix in the Chinese page.

[imbajin]

🧹 Minor: The standalone installation above leaves the shell in hugegraph-ai/vermeer-python-client, so this command resolves to a nonexistent nested vermeer-python-client/vermeer-python-client/... path. Please either tell standalone users to run it from the repository root or use python src/pyvermeer/demo/task_demo.py there, and mirror the correction in the Chinese page.