apitest: speedup env termination by enabling zombie process reaping

* use `exec` when starting multideployer
* run services that struggle with reaping with dedicated PID 1 process (https://blog.phusion.nl/2015/01/20/docker-and-the-pid-1-zombie-reaping-problem/)

This two measures drops shutdown time from 10s to 4s.

localenv/docker-compose.yaml

@@ -28,6 +28,7 @@ services:
       ethereum: localhost:http://anvil:8545
       GRAPH_LOG: debug
       ETHEREUM_REORG_THRESHOLD: 0
+    init: true
 
   ipfs:
     image: ipfs/kubo:v0.27.0
@@ -74,6 +75,7 @@ services:
       IPFS_URL: http://ipfs:5001
       SUBGRAPH_ADMIN_URL: http://graph-node:8020
       SUBGRAPH_QUERY_URL: http://graph-node:8000
+    init: true
 
 networks:
   octant:

localenv/multideployer/entrypoint.sh

@@ -18,4 +18,4 @@ wait_for_rpc
 echo ""
 echo "anvil is ready!"
 
-python3 /app/server.py
+exec python3 /app/server.py