LedgerDB.V2: take snapshot at immutable tip

This ensures that the underlying handle isn't garbage-collected, as we only
schedule a garbage collection for all states with a slot number strictly smaller
than the immutable tip.

Also, a very small benefit is that we get (slightly) faster replay on node
startup.

Author: amesgen

ouroboros-consensus/src/ouroboros-consensus/Ouroboros/Consensus/Storage/LedgerDB/V2.hs

@@ -200,6 +200,7 @@ mkInternals bss h =
               TakeAtImmutableTip -> anchorHandle
               TakeAtVolatileTip -> currentHandle
           )
+            . volatileSuffix (ledgerDbCfgSecParam (ldbCfg env))
             <$> readTVarIO (ldbSeq env)
         Monad.void $
           takeSnapshot
@@ -387,6 +388,7 @@ implTryTakeSnapshot bss env mTime nrBlocks =
           (LedgerDBSnapshotEvent >$< ldbTracer env)
           (ldbHasFS env)
         . anchorHandle
+        . volatileSuffix (ledgerDbCfgSecParam (ldbCfg env))
         =<< readTVarIO (ldbSeq env)
       Monad.void $
         trimSnapshots