fix(daemon): propagate startup failure reason to user (GH#863)

When daemon fails to start due to legacy database or fingerprint validation,
the error was only logged to daemon.log. Users saw "Daemon took too long"
with no hint about the actual problem.

Changes:
- Write validation errors to .beads/daemon-error file before daemon exits
- Check for daemon-error file in autostart and display contents on timeout
- Elevate legacy database check in bd doctor from warning to error

Now when daemon fails due to legacy database, users see:
  "LEGACY DATABASE DETECTED!
   ...
   Run 'bd migrate --update-repo-id' to add fingerprint"

Instead of just "Daemon took too long to start".

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

cmd/bd/daemon.go

@@ -452,6 +452,12 @@ func runDaemonLoop(interval time.Duration, autoCommit, autoPush, autoPull, local
 	} else if err := validateDatabaseFingerprint(ctx, store, &log); err != nil {
 		if os.Getenv("BEADS_IGNORE_REPO_MISMATCH") != "1" {
 			log.Error("repository fingerprint validation failed", "error", err)
+			// Write error to daemon-error file so user sees it instead of just "daemon took too long"
+			errFile := filepath.Join(beadsDir, "daemon-error")
+			// nolint:gosec // G306: Error file needs to be readable for debugging
+			if writeErr := os.WriteFile(errFile, []byte(err.Error()), 0644); writeErr != nil {
+				log.Warn("could not write daemon-error file", "error", writeErr)
+			}
 			return // Use return instead of os.Exit to allow defers to run
 		}
 		log.Warn("repository mismatch ignored (BEADS_IGNORE_REPO_MISMATCH=1)")