fix: update example notebooks to use current graphrag_llm API

[octo-patch]

Fixes #2301

Problem

The example notebooks (local_search.ipynb, global_search.ipynb, drift_search.ipynb, global_search_with_dynamic_community_selection.ipynb) use removed/renamed APIs that no longer exist in the current codebase:

• from graphrag.config.enums import ModelType → ModelType was removed from enums.py
• from graphrag.config.models.language_model_config import LanguageModelConfig → module does not exist
• from graphrag.language_model.manager import ModelManager → module does not exist
• from graphrag_vectors import LanceDBVectorStore → LanceDBVectorStore is not exported from the top-level graphrag_vectors package

These broken imports cause ImportError and NameError when users try to run the example notebooks.

Solution

Replace deprecated imports and model creation patterns with the current graphrag_llm API:

• ModelType + LanguageModelConfig → ModelConfig from graphrag_llm.config
• ModelManager().get_or_create_chat_model(...) → create_completion(config) from graphrag_llm.completion
• ModelManager().get_or_create_embedding_model(...) → create_embedding(config) from graphrag_llm.embedding
• get_tokenizer(config) → chat_model.tokenizer (tokenizer is already available on the model object)
• from graphrag_vectors import LanceDBVectorStore → from graphrag_vectors.lancedb import LanceDBVectorStore

These replacements align the notebooks with how the query/factory.py internally constructs models.

Testing

Verified the corrected imports match what is actually exported by the current graphrag_llm and graphrag_vectors packages in the repository.