Adopt pagination to list_shared_examples to ensure all data returned

[EugeneJinXin]

Bug and Reproduce

When using client.clone_public_dataset(), only the first 100 examples of a large dataset would be cloned instead of the full dataset. This causes data miss and user confusion.

Reproduction steps (attach a script at the end of PR as well):

1. Find a public dataset with >100 examples
2. Run client.clone_public_dataset(...)
3. Count examples in the cloned dataset
4. Result: Only 100 examples are copied

Solution

Refactored the method to use the existing helper function _get_paginated_list, which correctly handles paginated API responses.

This change resolves the input data extraction issue and is confirmed with a cross-langsmith site, see test section.

Test

With the PR's change, I conducted data clone (from langsmith personal cloud account, to a langsmith through local docker-compose) and successfully verified clone data. See reproduce with below script.

eujin@eujin-mn1 langsmith-project % python3 /Users/eujin/langsmith-project/main.py -v
Passed! All 214 rows cloned

def reproduce_issue():
    """ Clone a dataset with 200+ example, verify output is exact match"

    ls_client = Client(api_url='http://localhost:1980/api/v1')
    dataset_name = "eujin_test_200_rows"

    dataset_public_url = (
        "https://smith.langchain.com/public/0dfe83c3-079e-4ee3-b6a5-01a6508066ea/d"
    )
    ls_client.clone_public_dataset(dataset_public_url)
    cloned_dataset = ls_client.read_dataset(dataset_name=dataset_name)
    
    assert cloned_dataset.example_count == 214, f"Expected 214 examples, got {cloned_dataset.example_count}"
    
    print("Passed! All 214 rows cloned")

In addition, added a unit test in test_client.py to make sure list_shared_examples handles pagination correctly.

More Considerations
Feel free to let me know if you’d like to include more coverage in integration testing. I took a look of all methods using request_with_retries() in client.py, seems all safe (they either are POST, or single ID lookup with no pagination risk), however, i'm not sure about this one since i've not used splits feature, will verify offline with owner

[jacoblee93]

limit (Optional[int]): Maximum number of examples to return, by default None. -> limit (Optional[int]): Maximum number of examples to return. Defaults to no limit.