feat: Enable Trustless AI with Shade Agents and EZKL

[rainbowpuffpuff]

Motivation
Currently, data analysis and model training are performed in centralized environments, which requires trust in the platform operator. To build a truly decentralized and fair platform, we need a way to mathematically prove that the correct computations were performed on user data without revealing the data itself.

Proposed Solution
We will leverage a combination of Shade Agents and EZKL to create a robust and trustless system for processing sensitive data and running AI models.

• Shade Agents (for TEEs): We will use Shade Agents to run containerized Python scripts in a Trusted Execution Environment (TEE). This provides a secure and private environment for the computation.
• EZKL (for ZK-SNARKs): We will use EZKL to generate a Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (zk-SNARK) of the model's inference. This allows us to verifiably prove that a specific model was run on the data, without revealing the underlying data.

This combination gives us the best of both worlds: the privacy and security of TEEs for the computation, and the mathematical certainty of zk-SNARKs for the verification.

Use Cases:

• Fair Economic Model: Calculate Shapley values for data contributions in a verifiable way.
• Private AI Model Training: Train models on sensitive data (e.g., glucose data) and provide a verifiable proof of the model's execution.

Acceptance Criteria:

• A Shade Agent is deployed and capable of running a Python script in a TEE.
• EZKL is integrated to generate a zk-SNARK for a sample model.
• The results of the script execution and the ZK proof are recorded on-chain.

Alternatives Considered

• Shade Agents only: Using only TEEs provides a good level of security, but adding EZKL provides a stronger, mathematically verifiable guarantee of the computation's integrity.
• EZKL only: Using only EZKL is powerful for verification, but can be complex for general-purpose computation. Combining it with Shade Agents allows us to use the right tool for the right job.

Additional Context
This feature is a core component of the "Verifiable & Private AI" objective outlined in the project's TASKS.md file. It will serve as the epic for the more specific tasks related to Shade Agent and EZKL development.