feat(go): added DataPrompt and iterator-based stream support

[apascal07]

This PR adds Go iterator-based streaming support for model generation and prompt execution, enabling real-time consumption and parsing of model output chunks.

Usage:

GenerateStream / GenerateDataStream

Stream model responses using Go's range-over-func pattern:

// Untyped streaming - chunks are *ModelResponseChunk
for result, err := range genkit.GenerateStream(ctx, g,
    ai.WithPrompt("Tell me a joke"),
) {
    if err != nil { return err }
    if result.Done {
        fmt.Println("Final:", result.Response.Text())
    } else {
        fmt.Print(result.Chunk.Text())
    }
}

// Typed streaming - chunks are parsed to the output type
for result, err := range genkit.GenerateDataStream[*Recipe](ctx, g,
    ai.WithPrompt("Create a pasta recipe"),
) {
    if err != nil { return err }
    if result.Done {
        fmt.Printf("Recipe: %+v\n", result.Output)
    } else {
        fmt.Printf("Ingredients so far: %v\n", result.Chunk.Ingredients)
    }
}

DataPrompt

Strongly-typed prompts with Execute and ExecuteStream:

// Define a typed prompt
storyPrompt := genkit.DefineDataPrompt[StoryInput, *Story](g, "story",
    ai.WithPrompt("Write a story about {{topic}}"),
)

// Non-streaming execution returns typed output directly
output, resp, err := storyPrompt.Execute(ctx, StoryInput{Topic: "dragons"})
fmt.Printf("Title: %s\n", output.Title)

// Streaming execution
for result, err := range storyPrompt.ExecuteStream(ctx, StoryInput{Topic: "dragons"}) {
    if err != nil { return err }
    if result.Done {
        fmt.Printf("Title: %s\n", result.Output.Title)
    }
}

You can also wrap .prompt files with type information using LookupDataPrompt:

storyPrompt := genkit.LookupDataPrompt[StoryInput, *Story](g, "story")
output, resp, err := storyPrompt.Execute(ctx, StoryInput{Topic: "dragons"})