XL/objects: Initialize format.json outside of erasure. (#1640)

Fixes #1636

New format now generates a UUID and includes it along with
the order of disks. So that UUID is the real order of disks
and on command line user is able to specify disks in any order.

This pre-dominantly solves our dilemma.
```
{
   "format" : "xl",
   "xl" : {
      "version" : "1",
      "disk": "00e4cf06-5bf5-4bb5-b885-4b2fff4a7959",
      "jbod" : [
         "00e4cf06-5bf5-4bb5-b885-4b2fff4a7959",
         ....
         "c47d2608-5067-4ed7-b1e4-fb81bdbb549f",
         "a543293e-99f1-4310-b540-1e450878e844",
         "18f97cbe-529a-456a-b6d4-0feacf64534d"
      ]
   },
   "version" : "1"
}
```

object-common.go

@@ -22,19 +22,21 @@ import (
 )
 
 // Common initialization needed for both object layers.
-func initObjectLayer(storage StorageAPI) error {
+func initObjectLayer(storageDisks ...StorageAPI) error {
 	// This happens for the first time, but keep this here since this
 	// is the only place where it can be made expensive optimizing all
 	// other calls. Create minio meta volume, if it doesn't exist yet.
-	if err := storage.MakeVol(minioMetaBucket); err != nil {
-		if err != errVolumeExists {
-			return toObjectErr(err, minioMetaBucket)
+	for _, storage := range storageDisks {
+		if err := storage.MakeVol(minioMetaBucket); err != nil {
+			if err != errVolumeExists && err != errDiskNotFound {
+				return toObjectErr(err, minioMetaBucket)
+			}
+		}
+		// Cleanup all temp entries upon start.
+		err := cleanupDir(storage, minioMetaBucket, tmpMetaPrefix)
+		if err != nil {
+			return toObjectErr(err, minioMetaBucket, tmpMetaPrefix)
 		}
-	}
-	// Cleanup all temp entries upon start.
-	err := cleanupDir(storage, minioMetaBucket, tmpMetaPrefix)
-	if err != nil {
-		return toObjectErr(err, minioMetaBucket, tmpMetaPrefix)
 	}
 	return nil
 }