fix(smolagents): Reuse AnyLLM client instance to avoid event loop errors

[peteski22]

Summary

• Fixes the "Event loop is closed" error when making multiple completion calls with smolagents
• Creates an AnyLLM client instance once via AnyLLM.create() instead of returning the any_llm module
• Follows the same pattern used by tinyagent and openai framework implementations

Changes

• Parse model_id using AnyLLM.split_model_provider() to extract provider and model
• Store provider config separately from completion kwargs
• Create AnyLLM instance in create_client() for reuse by ApiModel base class
• Pass allow_running_loop=True to completion calls to permit sync API in async contexts
• Add error handling for malformed model_id with clear error message
• Add unit tests for AnyLLMModel class including regression test for issue smolagents AnyLLM client should be created only once and reused for completions #824

Test plan

• Existing unit tests pass (5 tests)
• New unit tests added (4 tests):
  • test_malformed_model_id_raises_clear_error
  • test_parses_model_id_correctly
  • test_create_client_creates_anyllm_instance
  • test_client_reused_across_multiple_calls (regression test)
• Pre-commit hooks pass (ruff, mypy, etc.)

Fixes #824

[codecov]

Codecov Report

❌ Patch coverage is 97.77778% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/any_agent/frameworks/smolagents.py	93.75%	1 Missing ⚠️

Files with missing lines	Coverage Δ
src/any_agent/testing/helpers.py	40.98% <ø> (-29.51%)	⬇️
tests/unit/frameworks/test_smolagents.py	100.00% <100.00%> (ø)
src/any_agent/frameworks/smolagents.py	61.83% <93.75%> (-9.48%)	⬇️

... and 43 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[HareeshBahuleyan]

Tested and worked with a sample script using smolagents with openai/gpt-4.1 model