feat: prefer stop over eos_token to align with openai finish_reason

[drbh]

This PR aligns the finish_reason value with openai. Currently TGI returns the string eos_token rather than stop as a finish_reason.

Fixes: #2296

The discrepancy between the values is easier to noticed when a strongly typed client is used. Currently the incorrect finish_value will cause the following request to fail since it expects stop rather than eos_token, these changes resolve this example.

use async_openai::config::OpenAIConfig;
use async_openai::types::CreateCompletionRequestArgs;
use async_openai::Client;
use std::error::Error;

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
    let client = Client::with_config(OpenAIConfig::new().with_api_base("http://localhost:3000/v1"));

    let request = CreateCompletionRequestArgs::default()
        .model("meta-llama/Meta-Llama-3-8B-Instruct")
        .prompt("What is Hugging Face and what does it do?")
        .max_tokens(2000_u32) // <- increased
        .seed(1337)
        .build()?;

    let response = client.completions().create(request).await?;

    println!("\nResponse (single):\n");
    for choice in response.choices {
        println!("{}", choice.text);
    }

    Ok(())
}

[ErikKaum]

this makes sense to me 👍