Add LLM config option for indexing stage and query stage

[tweinberger-lei]

Documentation recommends using only an instruct LLM (non-reasoning) for the indexing stage. However, using a reasoning LLM for the query stage could improve query results. Therefore, it would be great to support configuring different LLMs for the indexing and query stages in the config file, for example:

INDEXING_LLM_BINDING=openai
INDEXING_LLM_BINDING_HOST=https://api.openai.com/v1
INDEXING_LLM_BINDING_API_KEY=your_api_key
INDEXING_LLM_MODEL=gpt-5.4

QUERY_LLM_BINDING=openai
QUERY_LLM_BINDING_HOST=https://api.openai.com/v1
QUERY_LLM_BINDING_API_KEY=your_api_key
QUERY_LLM_MODEL=gpt-5-mini

Thanks!

[tweinberger-lei]

Similar to #2848

Looking forward to v1.5 :-)