test: make TEST-63 more reliable on slower machines

Otherwise we might never hit the trigger limit and wait indefinitely.

Found when trying to run the test on an EC2 xen machine without a nested
virt in CentOS CI (in preparations for some ... unforseseen consequences).

test/TEST-63-PATH/test.sh

@@ -7,17 +7,4 @@ TEST_DESCRIPTION="https://github.com/systemd/systemd/issues/17433"
 # shellcheck source=test/test-functions
 . "${TEST_BASE_DIR:?}/test-functions"
 
-test_append_files() {
-    (
-        # Tweak the trigger limit interval in case we are collecting coverage
-        # or running without KVM - in both cases we might be slow enough that
-        # we could miss the default rate-limit window and cause the test to fail
-        # unexpectedly.
-        if get_bool "$IS_BUILT_WITH_COVERAGE" || ! get_bool "$QEMU_KVM"; then
-            mkdir -p "${initdir:?}/etc/systemd/system/test63.path.d"
-            printf "[Path]\nTriggerLimitIntervalSec=10\n" >"${initdir:?}/etc/systemd/system/test63.path.d/triggerlimitinterval-override.conf"
-        fi
-    )
-}
-
 do_test "$@"

test/testsuite-63.units/test63.path

@@ -1,3 +1,6 @@
 # SPDX-License-Identifier: LGPL-2.1-or-later
 [Path]
 PathExists=/tmp/test63
+# Make the unit friendly to slower machines
+TriggerLimitIntervalSec=10
+TriggerLimitBurst=10