Fix: Add auto-flush to renumber, rename-prefix, and compact commands

These commands modify issues in bulk but weren't triggering auto-export
to JSONL. This caused database and JSONL to get out of sync.

Added markDirtyAndScheduleFlush() calls to:
- bd renumber (after renumbering completes)
- bd rename-prefix (after prefix rename completes)
- bd compact (after single/batch compaction)
- bd delete (already had it)

Fixes the issue where massive cleanups weren't exported to JSONL.

Amp-Thread-ID: https://ampcode.com/threads/T-a43dc9fa-e9bc-43c7-9055-33acc08bc642
Co-authored-by: Amp <amp@ampcode.com>

cmd/bd/compact.go

@@ -183,6 +183,9 @@ func runCompactSingle(ctx context.Context, compactor *compact.Compactor, store *
 		originalSize, compactedSize, savingBytes,
 		float64(savingBytes)/float64(originalSize)*100)
 	fmt.Printf("  Time: %v\n", elapsed)
+
+	// Schedule auto-flush to export changes
+	markDirtyAndScheduleFlush()
 }
 
 func runCompactAll(ctx context.Context, compactor *compact.Compactor, store *sqlite.SQLiteStorage) {
@@ -304,6 +307,11 @@ func runCompactAll(ctx context.Context, compactor *compact.Compactor, store *sql
 	if totalOriginal > 0 {
 		fmt.Printf("  Saved: %d bytes (%.1f%%)\n", totalSaved, float64(totalSaved)/float64(totalOriginal)*100)
 	}
+
+	// Schedule auto-flush to export changes
+	if successCount > 0 {
+		markDirtyAndScheduleFlush()
+	}
 }
 
 func runCompactStats(ctx context.Context, store *sqlite.SQLiteStorage) {

cmd/bd/rename_prefix.go

@@ -96,6 +96,9 @@ Example:
 			os.Exit(1)
 		}
 
+		// Schedule auto-flush to export changes
+		markDirtyAndScheduleFlush()
+
 		fmt.Printf("%s Successfully renamed prefix from %s to %s\n", green("✓"), cyan(oldPrefix), cyan(newPrefix))
 
 		if jsonOutput {

cmd/bd/renumber.go

@@ -127,6 +127,9 @@ Risks:
 			os.Exit(1)
 		}
 
+		// Schedule auto-flush to export changes
+		markDirtyAndScheduleFlush()
+
 		fmt.Printf("%s Successfully renumbered %d issues\n", green("✓"), len(issues))
 
 		// Count actual changes