I would have expected a generic warning that the command failed, maybe something more like:

if ! (eval "$SCRIPT"); then
    echo "Warning: Script failed (non-zero exit code)"
    if echo "$SCRIPT" | grep -q "sed"; then
        echo "Suggestion: did you use BSD sed syntax?"
        echo "Example: sed -i '' 's/foo/bar/' filename (Bad)"
    fi
fi

sed doesn't have an exit code for "error", so checking the exit code wouldn't work either way.

$ sed -i -e 's/FOOBAR/foobar/g' ./src/init.cpp && echo $?
0

"sed -i '' -e 's/FOOBAR/foobar/g' ./src/init.cpp && echo $?" results in no output for me since the sed command fails and the "echo $?" command is not executed but "sed -i -e 's/FOOBAR/foobar/g' ./src/init.cpp" does in fact work. I think this is because "sed -i -e 's/FOOBAR/foobar/g' ./src/init.cpp" is a legitimate sed command, it just doesn't cause any changes since FOOBAR is not present in the file.

In the OpenBSD 6.9 man pages for sed, it states EXIT STATUS The sed utility exits 0 on success, and >0 if an error occurs.

failing to replace is not an error

The line "git --no-pager diff --exit-code $commit && echo "OK" || (echo "Failed"; false) || RET=1" should check if the script failed to change anything, do you want to check each command and error if if has no change?

Do you want to not check for non zero exit codes but only check if the output of the script is the same as the commit?