Implement Langfuse Telemetry for generateImage

[toyamarinyon]

Currently, the generateImage function does not send telemetry data to Langfuse. While the AI SDK provides telemetry functionality, it is not yet implemented for generateImage (as of version 4.2.0).

This issue proposes implementing Langfuse telemetry specifically for generateImage. Langfuse supports image outputs (see https://langfuse.com/docs/tracing-features/multi-modality), allowing us to track and analyze the performance and characteristics of generated images.

Goal:

• Implement telemetry reporting to Langfuse when the generateImage function is executed.
• Ensure the telemetry data includes relevant information about the generated image, leveraging Langfuse's multi-modality support.

Considerations:

• This implementation will need to be independent of the existing AI SDK telemetry, as it's not yet supported for generateImage.
• Careful consideration should be given to the specific data points to include in the telemetry to provide meaningful insights into image generation performance.