Terminal closure leaves orphaned zsh subshell with high CPU in zsh <5.10

[LangLangBart]

to reproduce

Use the docker/podman command on the project's readme, including a --name flag
to connect more easily from another terminal.

podman run --name BUG_TEST  -e TERM -e COLORTERM \
  -e LC_ALL=C.UTF-8 -w /root -it --detach-keys="ctrl-^,ctrl-@" \
  --rm alpine sh -uec '
  apk add zsh curl tmux
  sh -c "$(curl -fsSL https://raw.githubusercontent.com/romkatv/zsh4humans/v5/install)"'

Options selected for z4h setup:

1   keyboard
1   keybindings
n   tmux
n   direnv
y   login shell

y   diamond
y   lock
... (prompt preferences do not matter for this bug, pick as you like)

Connect to 'BUG_TEST' in another terminal and run top to monitor processes.

podman exec -it BUG_TEST top

Go back to your original terminal where z4h was setup.

Run ls -l<TAB> and without making a selection from the fzf window. Press
ctrl-z while the fzf window is open until the fzf window with the selection
options is no longer visible. You may need to press ctrl-z twice.

Go back to the terminal window where the podman exec -it BUG_TEST top is
running. You should notice a zsh process that consumes a non-trivial amount of
CPU, in my case ~20%.

Mem: 566852K used, 1425028K free, 932K shrd, 2048K buff, 269824K cached
CPU:  19% usr   0% sys   0% nic  79% idle   0% io   0% irq   0% sirq
Load average: 0.95 0.49 0.20 2/212 988
  PID  PPID USER     STAT   VSZ %VSZ CPU %CPU COMMAND
  979   977 root     R     8420   0%   1  20% zsh -i
  980   979 root     T     696m  36%   2   0% /root/.cache/zsh4humans/v5/fzf/bin/fzf --query=^- --color=hl:201,hl+:201 --with-nth=2 --delimiter=\000 --ansi --exact --no-mouse --tie
  977     1 root     S     8420   0%   0   0% zsh -i
    1     0 root     S     8388   0%   3   0% zsh -i
  935     1 root     S     7132   0%   0   0% zsh -i
  936     1 root     S     7116   0%   0   0% zsh -i
  970     0 root     R     1708   0%   2   0% top

If you close all terminal windows and check your local CPU usage on your system,
you should notice a spike related to Docker/Podman processes until you run
podman kill BUG_TEST.

ps x -ro user,pid,pcpu,pmem,etime,comm | head -10

# USER    PID  %CPU %MEM  ELAPSED COMM
# hans  1400 100.1  4.1 12:03:32 /System/Library/Frameworks/Virtualization.framework/Versions/A/XPCServices/com.apple.Virtualization.VirtualMachine.xpc/Contents/MacOS/com.apple.Virtualization.VirtualMachine

---

I can consistently reproduce the z4h-fzf-complete bug on my MacBook using the
default .zshrc.mac file from zsh4humans as my zshrc with one extra line:

# easier to reproduce the issue with tmux off
zstyle ':z4h:' start-tmux false

When I set it up and opened Apple's default Terminal, entered ls -<TAB>, and
instead of pressing 'ctrl-z' just quit the entire terminal window, it always
spawns a single zsh process that continuously uses 100% CPU and runs
indefinitely until I manually kill it.

Added GIF to make the described issue clearer:

---

The issue is no longer reproducible with the zsh development version. I was able
to pinpoint it to the following commit where the issue no longer occurs:

• zsh-users/zsh@61610ea4

Is there a way to mitigate the issue until zsh devs release version 5.10?