Add lm studio as a LLM provider

[mpnikhil]

Added lmstudio as a LLM provider to be able to run GGUF models. With that we need to skip the tokens ourselves in the nodes where we expect structured response. Ollama json mode seems to do it for us but doesn't look like LMStudio API strips it away. I am still able to get structured json output outside of the block.

[MrZoidberg]

Any progress on merging this PR?

[rlancemartin]

Yes, will do now after resolving merge conflicts.