Fix non-zero exit code in load_dev_db script

Summary: load_dev_db.sh was failing. It turns out we try to end the kubectl port-forward process twice, once with a TERM signal and once with a KILL signal. In the event that the TERM signal works, the kill command will fail. This diff makes it so that we only escalate to a kill if the TERM didn't work.

Test Plan: ran the script, it now succeeds

Reviewers: vihang, michelle, philkuz

Reviewed By: michelle

Differential Revision: https://phab.corp.pixielabs.ai/D8459

GitOrigin-RevId: 7106408c5d71ca20c801b1e62a2f1351e7852ec0

scripts/load_dev_db.sh

@@ -90,5 +90,9 @@ fi
 # Kill kubectl port-forward.
 kill -15 "$!"
 sleep 2
-# Make sure process cleans up properly.
-kill -9 "$!" 2> /dev/null
+
+# Double check that it's dead.
+if pidof "$!"; then
+  kill -9 "$!"
+fi
+