Added Ollama for Easier access to LLM Models

[sidharthramesh]

Ollama - https://ollama.com/ provides an easy and quick way to run quantised models on consumer as well as GPU accelerated hardware. We would like to use some of the open-source models made available by Ollama with it's OpenAI APIs for the competition. Please allow us to add ollama to the runtime.

Also updated pip dependencies to install LanceDB

[chrisjkuch]

@sidharthramesh A few questions before accepting:

• Is there a reason the ollama python client library is not sufficient?

• Rather than invoking the general install script, can you pin the install to a specific version?

• Re: your comment

  We would like to use some of the open-source models made available by Ollama with it's OpenAI APIs for the competition.

  Please note that, as per the rules, only submissions that use open-source code libraries and models are eligible for prizes. Also note that the runtime does not have internet access so that even if OpenAI models were prize-eligible (which they are not), you cannot access the OpenAI API from the runtime.