Add Ollama-based RAG example

[ccreutzi]

Adapt OpenAI-based RAG example to Ollama.
Some minor code improvements carried back into the OpenAI example.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.04%. Comparing base (f8acb3a) to head (6f79fa2).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #45      +/-   ##
==========================================
- Coverage   97.11%   97.04%   -0.07%     
==========================================
  Files          38       39       +1     
  Lines        1246     1286      +40     
==========================================
+ Hits         1210     1248      +38     
- Misses         36       38       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[MiriamScharnke]

This is a really good example! Clear story, great title, shows a useful workflow! I have some stylistic feedback, but I'm hitting "approve" anyway, because it's definitely reached the level of review I've been able to give the other examples!

Feedback:

• Opening paragraph, when I open the example, I see "-> title "Generate Response" " - is this meant to be a link? If not, remove.
• Consolidate requirements into a single paragraph: To run this example, you need Text Analytics Toolbox. You also need a valid API key etc.

Download and Preprocess Documents:

• add opening sentence: "The document corpus used in this example contains reports that ". Download the data.
  (then add url=[etc] in the same code block as the next bit)

• generally, use Name=Value instead of 'Name',Value syntax

• Define the function to read the text -> Create a file datastore from the downloaded files.

• if possible, use @extractFileText directly instead of setting up a separate function handle.

• helper function preprocessDocuments -> preprocessDocuments function, defined at the end of this example.

• put the line initializing the chatbot into the next section, or its own section - it doesn't really belong to the data preprocessing step.

Retrieve Relevant Documents:

• link to TAT functions ref pages when possible, such as bm25Similarity
• "Iterate over sorted document indices until word limit is reached" -> keep using full sentences! Also explain what the iteration is doing.

[ccreutzi]

Talked to Miriam, and we agreed to move the recommended changes into the future (and probably onto her plate).