Title: st.progress() / other widgets raise AttributeError: 'NoneType' object has no attribute 'get_name' when called from a Python callback invoked by JS, even after add_script_run_ctx

[christykanto]

When a Python callback is invoked asynchronously from JS (e.g. via pyodide.ffi.create_proxy, the pattern used to receive progress callbacks from a JS library like WebLLM), calling any Streamlit widget (st.progress, st.empty().caption, etc.) from inside that callback raises:

AttributeError: 'NoneType' object has no attribute 'get_name'

at streamlit/runtime/scriptrunner_utils/script_run_context.py, inside get_script_run_ctx().

I followed the documented workaround for calling Streamlit from background threads (capturing get_script_run_ctx() beforehand and calling add_script_run_ctx(threading.current_thread(), ctx) inside the callback), but it still fails at the same line - I suspect threading under Pyodide doesn't behave the way Streamlit's context-tracking expects (no real OS threads).

Minimal reproducer:
import asyncio
import streamlit as st
from streamlit.runtime.scriptrunner import add_script_run_ctx, get_script_run_ctx

st.title("Minimal repro: st.progress() inside a JS-invoked callback")

if st.button("Run"):
progress_bar = st.progress(0)
ctx = get_script_run_ctx()

def on_tick(value):
    add_script_run_ctx(__import__("threading").current_thread(), ctx)
    progress_bar.progress(value)

async def fake_async_work():
    import js  # type: ignore[import-not-found]
    from pyodide.ffi import create_proxy

    proxy = create_proxy(on_tick)
    # Simulate a JS library invoking our Python callback asynchronously,
    # the same way WebLLM's initProgressCallback does.
    await js.eval("new Promise(r => setTimeout(r, 50))")
    proxy(0.5)
    await js.eval("new Promise(r => setTimeout(r, 50))")
    proxy(1.0)
    proxy.destroy()

asyncio.run(fake_async_work())

Expected: the widget updates using the captured context, same as the documented threading pattern works in server-based Streamlit.

Actual: crashes with the AttributeError above, even with add_script_run_ctx.

Use case: wiring up initProgressCallback from @mlc-ai/web-llm (via import js + create_proxy) to show real download progress for an in-browser LLM - this is a common pattern for any JS library with an async progress callback.