test: add unit tests for llm/utils module

[Jeong-Ryeol]

Following up on @0xallam's comment on #157:

"I think we need to add even more unit tests now, especially to the most critical parts of the agent."

This PR adds tests for llm/utils.py - the core LLM response parsing logic.

Addresses #158

What's included

Test Class	Coverage	Tests
TestParseToolInvocations	XML tool invocation parsing	8
TestFixStopword	Incomplete function tag handling	4
TestFormatToolCall	Tool call XML formatting	4
TestCleanContent	Content cleaning and XML removal	6
TestTruncateToFirstFunction	First function truncation	4

Total: 26 tests, all passing.

Test plan

• All 26 tests pass locally
• Follows existing test patterns from test_argument_parser.py
• Covers edge cases (empty content, HTML entities, multiline values, etc.)

Next steps (if this direction looks good)

I'm planning to continue adding tests for other critical modules:

Module	Why it's critical
agents/state.py	Agent execution state management
llm/memory_compressor.py	Context window optimization
tools/executor.py	Tool execution logic

Happy to adjust priorities based on your feedback!