fix: improve graph loading for evals

[cristipufu]

Description

Optimize evaluation runs by loading the LangGraph configuration once and reusing it across all evaluation executions, reducing redundant initialization and improving performance.

Ref: #190

Development Package

• Add this package as a dependency in your pyproject.toml:

[project]
dependencies = [
  # Exact version:
  "uipath-langchain==0.0.139.dev1002090661",

  # Any version from PR
  "uipath-langchain>=0.0.139.dev1002090000,<0.0.139.dev1002100000"
]

[[tool.uv.index]]
name = "testpypi"
url = "https://test.pypi.org/simple/"
publish-url = "https://test.pypi.org/legacy/"
explicit = true

[tool.uv.sources]
uipath-langchain = { index = "testpypi" }

[Copilot]

Pull Request Overview

This PR improves the graph loading mechanism for evaluations by refactoring the runtime context management to use an async context manager pattern. The changes restructure the code to use a resolver-based approach for graph loading instead of the previous script-based runtime.

• Refactored runtime creation to use LangGraphJsonResolverContext for async graph loading
• Improved error handling with full traceback logging
• Restructured async function execution flow within the middleware

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
src/uipath_langchain/_cli/cli_eval.py	Main refactoring to use async context manager for graph resolver and improved error handling
pyproject.toml	Version bump from 0.0.138 to 0.0.139

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}