feat: Add warnings for invalid tool_choice and UTs

[CatherineSue]

Motivation

This PR improves error handling and user experience for invalid tool choices in SGLang's function calling system. Previously, when users specified a tool_choice with a function name that doesn't exist in the available tools, the system would silently fail with Skip invalid EBNF strings. This change adds proper validation and logging.
NOTE: the correct way is to add this kind of validation in the ChatCompletionRequest protocol and returns a 400. However, our adapter.py or protocol.py doesn't have this kind of check. And it is better to fix all similar issues in refactor. So right now we use a logging.warning to mitigate it.

The changes also introduce comprehensive unit tests for tool choice functionality across different models and scenarios, ensuring robust validation of the function calling system.

Modifications

• Add validation in get_ebnf() method to check if the requested function exists in available tools
• Add test_tool_choice.py to run_suite:
  • tool_choice="auto" in streaming and non-streaming modes
  • tool_choice="required" in streaming and non-streaming modes
  • Specific function selection via tool_choice={"type": "function", "function": {"name": "..."}}
  • Multi-tool scenarios
  • Error handling for invalid function names
  • Support for multiple model types (Llama, Qwen2.5, Mistral), which covers (Llama32Detector, MistralDetector, and Qwen25Detector)

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://slack.sglang.ai to discuss your PR.

[CatherineSue]

UT result:
python3 -m unittest test_tool_choice.TestToolChoice

[CatherineSue]

Updated UT with flaky tests instead of print warnings

[JustinTong0323]

LGTM.

[zhyncs]

verified locally