xl: Rewrite auto-healing and implement auto new-disk healer (#8114)

The new auto healing model selects one node always responsible
for auto-healing the whole cluster, erasure set by erasure set.
If that node dies, another node will be elected as a leading
operator to perform healing.

This code also adds a goroutine which checks each 10 minutes
if there are any new unformatted disks and performs its healing
in that case, only the erasure set which has the new disk will
be healed.

cmd/admin-heal-ops.go

@@ -53,6 +53,10 @@ const (
 	// time-duration to keep heal sequence state after it
 	// completes.
 	keepHealSeqStateDuration = time.Minute * 10
+
+	// nopHeal is a no operating healing action to
+	// wait for the current healing operation to finish
+	nopHeal = ""
 )
 
 var (
@@ -548,7 +552,7 @@ func (h *healSequence) queueHealTask(path string, healType madmin.HealItemType)
 	var respCh = make(chan healResult)
 	defer close(respCh)
 	// Send heal request
-	globalBackgroundHealing.queueHealTask(healTask{path: path, responseCh: respCh, opts: h.settings})
+	globalBackgroundHealRoutine.queueHealTask(healTask{path: path, responseCh: respCh, opts: h.settings})
 	// Wait for answer and push result to the client
 	res := <-respCh
 	if !h.reportProgress {
@@ -587,6 +591,8 @@ func (h *healSequence) healItemsFromSourceCh() error {
 
 		var itemType madmin.HealItemType
 		switch {
+		case path == nopHeal:
+			continue
 		case path == SlashSeparator:
 			itemType = madmin.HealItemMetadata
 		case !strings.Contains(path, SlashSeparator):