Avoid loading cluster log at startup

This fixes a problem with datalad's test suite, where loading the cluster
log happened to cause the git-annex branch commits to take a different
shape, with an additional commit.

It's also faster though, since many commands don't need the cluster log.

Just fill Annex.clusters with a thunk.

Sponsored-by: the NIH-funded NICEMAN (ReproNim TR&D3) project

Logs/Cluster.hs

@@ -11,6 +11,7 @@ module Logs.Cluster (
 	module Types.Cluster,
 	getClusters,
 	loadClusters,
+	preLoadClusters,
 	recordCluster,
 ) where
 
@@ -24,7 +25,12 @@ import qualified Data.Map as M
 import qualified Data.Set as S
 
 getClusters :: Annex Clusters
-getClusters = maybe loadClusters return	=<< Annex.getState Annex.clusters
+getClusters = maybe loadClusters id =<< Annex.getState Annex.clusters
+
+{- This works around a module dependency loop. -}
+preLoadClusters :: Annex ()
+preLoadClusters = Annex.changeState $ \s ->
+	s { Annex.clusters = Just loadClusters }
 
 {- Loads the clusters and caches it for later.
  -
@@ -37,5 +43,5 @@ loadClusters = do
 	dead <- (S.fromList . map ClusterNodeUUID)
 		<$> trustGet DeadTrusted
 	clusters <- getClustersWith (M.map (`S.difference` dead))
-	Annex.changeState $ \s -> s { Annex.clusters = Just clusters }
+	Annex.changeState $ \s -> s { Annex.clusters = Just (pure clusters) }
 	return clusters