fix: allow danging objects to be purged properly deleteMultipleObjects() (#14273)

Deleting bulk objects had an issue since the relevant versionID
is not passed through the layers to ensure that the dangling
object purge actually works cleanly.

This is a continuation of quorum related error returned by
multi-object delete API from #14248

This PR ensures that we pass down correct information as
well as extend the scope of dangling object detection.

cmd/erasure-server-pool.go

@@ -335,7 +335,9 @@ func (z *erasureServerPools) getPoolIdxExistingWithOpts(ctx context.Context, buc
 			pinfo := poolObjInfo{
 				PoolIndex: i,
 			}
-			opts.VersionID = "" // no need to check for specific versionId
+			// do not remove this check as it can lead to inconsistencies
+			// for all callers of bucket replication.
+			opts.VersionID = ""
 			pinfo.ObjInfo, pinfo.Err = pool.GetObjectInfo(ctx, bucket, object, opts)
 			poolObjInfos[i] = pinfo
 		}(i, pool, poolOpts[i])
@@ -353,15 +355,15 @@ func (z *erasureServerPools) getPoolIdxExistingWithOpts(ctx context.Context, buc
 	})
 
 	for _, pinfo := range poolObjInfos {
-		if pinfo.Err != nil && !isErrObjectNotFound(pinfo.Err) {
-			return -1, pinfo.Err
-		}
-
 		// skip all objects from suspended pools for mutating calls.
 		if z.IsSuspended(pinfo.PoolIndex) && opts.Mutate {
 			continue
 		}
 
+		if pinfo.Err != nil && !isErrObjectNotFound(pinfo.Err) {
+			return -1, pinfo.Err
+		}
+
 		if isErrObjectNotFound(pinfo.Err) {
 			// No object exists or its a delete marker,
 			// check objInfo to confirm.