refactor: go back to the llama3 model

FACT: Improving comments

fix: Implement robust post-processing to eliminate LLM dialogue hallucination

fix: Implement output cleaning for conversational hallucinations

fix: Final stop sequence added to Modelfile to halt LLM hallucination

fix: Stricter prompt engineering in Modelfile to prevent LLM hallucination

refactor: Switch LLM from Llama 3 to custom 'ishu-companion' model

Delete __pycache__/assistant.cpython-313.pyc

feat(llm/multitool): Implement batched tool call execution

fix(llm/behavior): Force conversational output after tool execution

fix(llm/tooling): Enhance Ollama stability and tool execution

test(routine): Add comprehensive tests for routine modification (add/remove)

fix(scheduler): Correctly identify next task when tasks are back-to-back

feat(scheduler): Refine time logic and implement local NOW/NEXT task distinction

FIX: fix the path of the routine.json