Easier evaluations with LangSmith SDK v0.2

LangSmith has released v0.2 of their SDKs with improvements to the developer experience for evaluating applications. These include simplified usage of evaluate()/aevaluate() methods, an option to run evaluations locally without uploading results, improved SDK performance, and expanded documentation. The changes have been made in both Python and TypeScript SDKs. Two breaking changes were introduced in the Python SDK: a default max_concurrency value change from None to 0, and string data argument is now checked for UUID before being treated as dataset name.