Fixes flagging when allow_flagging is set to "auto"

[abidlabs]

Turns out we didn't implement allow_flagging set to "auto" when we re-implemented Interface to be constructed from Blocks. This restores that functionality.

Test with a script like this:

import os
import gradio as gr

def calculator(num1, operation, num2):
    if operation == "add":
        return num1 + num2
    elif operation == "subtract":
        return num1 - num2
    elif operation == "multiply":
        return num1 * num2
    elif operation == "divide":
        return num1 / num2

iface = gr.Interface(
    calculator,
    ["number", gr.Radio(["add", "subtract", "multiply", "divide"]), "number"],
    "number",
    allow_flagging="auto",
)

iface.launch()

Fixes: #2603

After discussion with @aliabid94, modified the functionality of "auto" to only log the input data. This simplifies the implementation considerably because there's no easy way to flag the data "once the prediction has completed running", and we think approach is reasonable for this relatively rare use case. Updated the documentation to reflect this functionality

[github-actions]

All the demos for this PR have been deployed at https://huggingface.co/spaces/gradio-pr-deploys/pr-2695-all-demos

[aliabid94]

LGTM

[abidlabs]

Thanks for the review @aliabid94!