Add infrastructure skeleton for review (refs #21)

Lays out the agreed shape from the design discussion: one flake at the
root with three nixosConfigurations (mqtt-ingest, db-host, app-host),
vendored Supabase docker-compose pointed at the external db-host,
Caddy → Kong → PostgREST/GoTrue, a Go MQTT→Postgres ingester stub,
deploy-rs node map, and a Gitea Actions workflow that deploys in
db → app → ingest order on push to main.

No real implementation yet — every host module has TODOs marking the
gaps (real hardware config, actual hostnames, HiveMQ packaging,
vendor the full upstream Supabase compose, etc.).

nix/dev/process-compose.nix

@@ -0,0 +1,30 @@
+{ pkgs, services-flake }:
+
+{ ... }:
+{
+  imports = [ services-flake.processComposeModules.default ];
+
+  # Skeleton — services-flake gives us postgres, plus we'd add Supabase via
+  # docker-compose run, Caddy, Kong (or skip Kong locally and hit services
+  # directly), the ingester, and `vite dev`. Fill in as we go.
+  services = {
+    postgres."db" = {
+      enable = true;
+      port = 5432;
+      initialDatabases = [{ name = "gebos"; }];
+    };
+  };
+
+  settings.processes = {
+    ingester = {
+      command = "${pkgs.go}/bin/go run ./ingester";
+      depends_on."db".condition = "process_healthy";
+    };
+    frontend = {
+      command = "${pkgs.nodejs_20}/bin/npm --prefix frontend run dev";
+    };
+    # TODO: supabase compose (point POSTGRES_HOST at the services-flake db)
+    # TODO: kong (use the vendored kong.yml from nix/supabase/)
+    # TODO: caddy (mirror prod routes locally, no TLS)
+  };
+}