InterpreterPoolExecutor doesn't respect "thread_name_prefix"

[paultiq]

Bug report

Bug description:

InterpreterPoolExecutor takes a thread_name_prefix parameter, but it doesn't propagate to the underlying subinterpreter.

from concurrent.futures import ThreadPoolExecutor
try:
 from concurrent.futures.interpreter import InterpreterPoolExecutor # 3.14+
except ModuleNotFoundError:
 from interpreters_backport.concurrent.futures.interpreter import InterpreterPoolExecutor 

with ThreadPoolExecutor(thread_name_prefix="test_tpe") as tpe_executor:
 tpe_executor.submit(exec, "import threading;print(threading.current_thread().name)") # 
 
with InterpreterPoolExecutor(thread_name_prefix="test_ipe") as ipe_executor:
 ipe_executor.submit(exec, "import threading;print(threading.current_thread().name)")

Produces

test_tpe_0
Dummy-1

CPython versions tested on:

CPython main branch

Operating systems tested on:

Windows

[ZeroIntensity]

Confirmed on the main branch, but this isn't related to InterpreterPoolExecutor. ipe_executor._thread_name_prefix does appear to be set to the correct value, and the Thread object name is the right value when checking it directly with ipe_executor._threads.

The problem is that thread names aren't pushed down to subinterpreters, which very well might be a wontfix depending on the complexity of the change. Here's a reproducer for that:

import _interpreters
import threading

source = """
import threading

print(threading.current_thread().name) # Dummy-1
"""

def main():
 interp = _interpreters.create()
 print(threading.current_thread().name) # Hello
 _interpreters.run_string(interp, source)

threading.Thread(target=main, name="hello").start()

cc @ericsnowcurrently, it might be better to just note this when subinterpreters get documented better. (Though, if that's the case, thread_name_prefix should just get removed from InterpreterPoolExecutor.)