fix(sling): use correct beads database for rig-level beads (gt-n5gga)

When slinging rig-level beads (gt-*, bd-*, etc.), the BEADS_DIR was
unconditionally set to town beads, which could bypass the redirect-based
routing needed for these beads. This caused assignee updates to potentially
fail silently or target the wrong database.

Changes:
- sling.go: Only set BEADS_DIR for town-level (hq-*) beads; rig-level
  beads now use redirect from polecat worktree for proper routing
- convoy.go: Add --no-daemon to bd show calls to ensure fresh data

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

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

internal/cmd/convoy.go

@@ -1130,8 +1130,9 @@ func getIssueDetailsBatch(issueIDs []string) map[string]*issueDetails {
 		return result
 	}
 
-	// Build args: bd show id1 id2 id3 ... --json
-	args := append([]string{"show"}, issueIDs...)
+	// Build args: bd --no-daemon show id1 id2 id3 ... --json
+	// Use --no-daemon to ensure fresh data (avoid stale cache from daemon)
+	args := append([]string{"--no-daemon", "show"}, issueIDs...)
 	args = append(args, "--json")
 
 	showCmd := exec.Command("bd", args...)
@@ -1177,7 +1178,8 @@ func getIssueDetailsBatch(issueIDs []string) map[string]*issueDetails {
 // Prefer getIssueDetailsBatch for multiple issues to avoid N+1 subprocess calls.
 func getIssueDetails(issueID string) *issueDetails {
 	// Use bd show with routing - it should find the issue in the right rig
-	showCmd := exec.Command("bd", "show", issueID, "--json")
+	// Use --no-daemon to ensure fresh data (avoid stale cache)
+	showCmd := exec.Command("bd", "--no-daemon", "show", issueID, "--json")
 	var stdout bytes.Buffer
 	showCmd.Stdout = &stdout