Where is the output of an aborted shell command supposed to be found?

Hi,

with GNU Emacs 30.2 (Fedora 43), I've stumbled upon some
behaviour that I do not understand at first (and second)
glance.  Consider a shell script stderr-test.sh:

| #!/bin/sh

| /bin/echo Test for stderr. >&2
| /bin/sleep "$1"
| exit 1

(/bin/echo so that it flushes its stderr when finishing.)

If I run M-! sh stderr-test.sh 0 RET, I observe that:

- The echo area displays "Test for stderr.",

- "Test for stderr." is appended to *Messages*, and

- the buffer *Shell Command Output* is created, has the
  "mode" "Fundamental - Exit [1]" and contains "Test for
  stderr."

If I run M-! sh stderr-test.sh 999 RET and abort the process
with C-g:

- The echo area displays "Quit",

- *Messages* is appended with:

  | Waiting for process to die...done
  | (Shell command failed with error)
  | Quit

- and the *Shell Command Output* buffer is empty, but has
  the "mode" "Fundamental - Error".

When I repeat the same test for stdout, it shows the same
results.

My expectation is that when I abort a shell command, *Shell
Command Output* contains all output up to the point of
killing the process (plus/minus buffering), and reading the
docstring for shell-command does not immediately refute
this.

Is the output of an aborted shell command silently
discarded, or is it stored somewhere else?

TIA,
Tim