Generative AI drivers + AWS Bedrock driver implementation

[clpetersonucf]

Ready for review.

• Transitions generative AI tooling to be more aligned with the user asset driver model.
• Created a boto3 session service to enable boto3 client instantiation for S3 and Bedrock.
• Created a GenerationCore class for driver-agnostic generation operations (like prompt assembly) and GenerationDriver abstract class that serves as the baseline for individual drivers.
• Created a GenerationDriverFactory class to manage dynamic loading of the specified generation driver.
• Moved django configurations around a bit. There is now an aws.py settings file with generalized AWS configurations like profile, account ID and secret, and credential provider. When applicable these will fall back to the S3-specific environment configs.
• boto3 session configuration now supports AWS profiles, with a fallback to account ID/secret values if not set.
• Generation drivers (at least Bedrock) support both query_sync and query_streaming interfaces for use with InvokeModel and InvokeModelWithResponseStream respectively.
• Added a new /api/generation/streaming/ endpoint to facilitate response streaming.
• Added engine core promptResponse and promptRejection postMessage hooks.
• Added engine core promptStreamingResponse and promptStreamingRejection to enable response streaming handling in widget players.