Memory usage over time (many prompts via inquirerpy)

Hi,
thanks a lot for this useful package.

Originally raised on inquirerpy (a prompt-toolkit wrapper); author advised to submit here instead:

I'm building an interactive app (with inquirerpy, not using asyncio) involving large lists and jumping back and forth instantiating new prompts on the way; it is intended to run for a while.
I noticed memory usage keeps growing and very little being released over time.

Am I right in assuming on event.app.exit() the prompt should exit and tear down entirely? Is there anything else I can/should do to reset/clean up?

Python 3.10.2 on windows (tags/v3.10.2:a58ebcc, Jan 17 2022, 13:59:55) [MSC v.1929 32 bit (Intel)]
inquirerpy==0.3.3
prompt-toolkit==3.0.28

Here's a simple demo to highlight the issue:

import time, random
from InquirerPy import inquirer
from pynput.keyboard import Key, Controller
from mem_top import mem_top

def makeChoices():
	for i in range(1000):
		words = "The value that should be presented to the user prior/after selection of the choice. This value is optional, if not provided, it will fallback to the string representation of value.".split()
		random.shuffle(words) # not needed to reproduce
		yield " ".join(words)

def getChoices():
	choices = []
	for choice in makeChoices(): choices.append(choice)
	return choices

def _prompt(i):
	inquirer.select(message="Testmem prompt" if i > 0 else "Press enter to start test", choices=getChoices(),
		default=None, show_cursor=False).execute()
	if (i < 100): # this runs in the next iteration
		for i in range(1000):
			keyboard.press(Key.down)
			keyboard.release(Key.down)
		keyboard.press(Key.enter)
		keyboard.release(Key.enter)

keyboard = Controller()

i = -1
while i < 100:
	i += 1
	time.sleep(0.5)
	_prompt(i)

print(mem_top())

"""
refs:
9093    <class 'prompt_toolkit.filters.base._AndCache'> {(~Condition(<function is_done at 0x03BE0FA0>), Condition(<function BaseComplexPrompt.__init__.<loca...
4001    <class 'prompt_toolkit.formatted_text.base.FormattedText'> FormattedText([('class:pointer', '❯'), ('class:marker', ' '), ('[SetCursorPosition]', ''), ('class:p...

types:
202428   <class 'dict'>
169534   <class 'prompt_toolkit.key_binding.key_bindings.Binding'>
155551   <class 'tuple'>
118098   <class 'list'>
59785    <class 'function'>
35636    <class 'frozenset'>
34023    <class 'prompt_toolkit.styles.base.Attrs'>
27188    <class 'cell'>
8699     <class 'collections.deque'>
8389     <class 'prompt_toolkit.cache.SimpleCache'>
"""

By the way, a feasible workaround is using multiprocessing for the prompts, however I'd prefer keeping things elegant and snappy.

Start by running the supplied Python 3.10 demo with inquirerpy and prompt-toolkit 3.0.28, then inspect the lifecycle around inquirer.select(...).execute() and event.app.exit(). Use the listed retained types and references as leads for determining what survives each prompt; done means identifying the retention cause and confirming memory is released or documenting the required cleanup.