fix(consul-postgres-ha): pass --base-id to each Envoy to avoid hot-restart socket clash

Both Envoys (webdemo + postgres) on a worker default to base-id=0 for
their hot-restart Unix domain socket. The second one to start hits
errno=98 (EADDRINUSE) and exits in a loop. Fixed by passing
`--base-id 1` (webdemo) and `--base-id 2` (postgres). Distinct base
IDs use distinct domain socket paths.

Visible symptom before this fix: ps showed only one envoy process
running on workers (whichever started first). The other's supervise
loop spun on `unable to bind domain socket with base_id=0, id=0,
errno=98`. postgres-master upstream listener (127.10.0.20:5432) was
never bound, so Patroni replicas got "connection refused" on
pg_basebackup.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: h4x3rotab

consul-postgres-ha/mesh-sidecar/entrypoint.sh

@@ -236,7 +236,10 @@ EOF
             -admin-bind="127.0.0.1:19000" \
             -bootstrap \
             > /tmp/envoy-webdemo.json 2>/dev/null; then
-        envoy -c /tmp/envoy-webdemo.json -l info 2>&1
+        # --base-id distinguishes this envoy from the postgres one
+        # below; otherwise both fight over the default base-id=0
+        # hot-restart domain socket and one fails with EADDRINUSE.
+        envoy -c /tmp/envoy-webdemo.json -l info --base-id 1 2>&1
       else
         echo "[envoy-webdemo] waiting for webdemo sidecar registration..."
       fi
@@ -255,7 +258,8 @@ EOF
             -admin-bind="127.0.0.1:19001" \
             -bootstrap \
             > /tmp/envoy-postgres.json 2>/dev/null; then
-        envoy -c /tmp/envoy-postgres.json -l info 2>&1
+        # See note on --base-id in the webdemo envoy block above.
+        envoy -c /tmp/envoy-postgres.json -l info --base-id 2 2>&1
       else
         echo "[envoy-postgres] waiting for ca leaf / proxy registration..."
       fi