embedder: added OpenAIGenerator model implementation

[jordanh]

Issue - Enhancement

The "research" version of the embedder service was aimed at enabling local development against AI and GPT models. For production, we're likely to utilize Open AI for summarization of textual objects to embed in much the same manner as how we use Open AI for creating meeting summaries, template recommendations, etc.

Acceptance Criteria (optional)

We can:

• Enable Open AI for summarization in a .env file via a configuration string similar to:

  AI_GENERATION_MODELS='[{"model": "openai:gpt-4"}]'

• Engineer and test a summarization prompt that works well for extant fullText transforms (i.e. retrospectiveDiscussionTopic)

Estimated effort: 13 points (see CONTRIBUTING.md)