Record SMTP deployment blocker and remaining restart (#57)

docs/signup-facts.md

@@ -124,7 +124,23 @@ MongoDB container (90-second context timeout). No checks were disabled. Fresh
 stable image pins resolved on retry; two image builds hit a Go proxy TLS
 handshake timeout during protobuf generator installation. The generator stage
 now shares the existing BuildKit Go module/build caches, preserving its pinned
-tool versions and generation command; the image rebuild is in progress.
+tool versions and generation command. Cache seeding also stalled during file
+extraction; it was canceled after copying the pinned module archives. A normal
+build retry remained blocked before starting a compiler process. A fallback
+using the same freshly pulled, digest-pinned Go image, canonical Make generation
+command and existing module cache stalled during ordinary container startup,
+before package installation or compilation produced output.
+
+Deployment is not complete: the running notifications container still uses the
+previous image. The SMTP Vault password and source configuration are prepared,
+but its Agent policy/template have not been activated against that old image.
+Vault responds over HTTP, while Vault/Mongo Docker health checks time out and
+the local Sendico session is unavailable. Restarting Docker Desktop was requested
+for recovery because it also interrupts the separate Sendico stack; it has not
+been authorized or performed. Rebuild, activate the matching Agent configuration,
+restart notifications and verify the live `email_driver: smtp` startup log after
+Docker is responsive. Do not treat the successful SMTP login as a deployed
+notification delivery check.
 
 
 With `api.sendico` configured, the product BFF creates the local pending customer