ReasoningBank for Agents using Analytics plugin

### Background Context

The BigQuery-Agent-Analytics plugin provides users with access to valuable data that can significantly improve agent performance and customer support quality. A primary application for this data involves establishing a "ReasoningBank" based on previous agent experiences. By distilling historical issue trajectories into concise memories, relevant insights can be injected into the agent's context to assist with similar future tasks.

### Action plan

Develop a ReasoningBank solution that users can activate via their plugin configuration. Key design and implementation steps include:

1. Creating a BigQuery schema dedicated to storing distilled memories.
2. Formulating a process to distill interaction trajectories into actionable memories.
3. Building a RAG system to identify and fetch appropriate memories for specific tasks.
4. Creating a mechanism for context injection.
5. Enabling user customization for LLM selection, text embedding models, and memory retrieval prompts.

### Primary Objectives

Leverage the BigQuery-Agent-Analytics plugin to upgrade agent capabilities by:

1. Improving the overall success rate of issue resolution.
2. Minimizing the number of steps required to reach a successful outcome.