go/transaction: Fix Abort to wait for synchronizers completion

There was a thinko in transaction.Abort - it was spawning synchronizers
under a waitgroup, but wg.Wait() call was forgotten. This way, e.g. in
ZODB if a transaction was aborted, corresponding connection could be not
yet returned back into DB pool.

Fix it.

Test is TODO for the time when, hopefully, tracetest is generally ready.

go/transaction/transaction.go

-// Copyright (C) 2018  Nexedi SA and Contributors.
+// Copyright (C) 2018-2019  Nexedi SA and Contributors.
 //                          Kirill Smelkov <kirr@nexedi.com>
 //
 // This program is free software: you can Use, Study, Modify and Redistribute
@@ -160,6 +160,7 @@ func (txn *transaction) Abort() {
 			syncv[i].AfterCompletion(txn)
 		}()
 	}
+	wg.Wait()
 
 	// XXX return error?
 }

go/transaction/transaction_test.go

-// Copyright (C) 2018  Nexedi SA and Contributors.
+// Copyright (C) 2018-2019  Nexedi SA and Contributors.
 //                          Kirill Smelkov <kirr@nexedi.com>
 //
 // This program is free software: you can Use, Study, Modify and Redistribute
@@ -98,7 +98,17 @@ func TestAbort(t *testing.T) {
 	dm.Modify()
 
 	// XXX +sync
+/* XXX test Abort not waiting for AfterCompletion via tracetest (with injected δt shakes):
+	txn.RegisterSync(AfterCompletion: {eAfterCompletion})
 
+	go func() {
+		eAbortPre
+		txn.Abort()
+		eAbortPost
+	}()
+
+	expect(eAbortPre, eAfterCompletion, eAbortPost)
+*/
 	txn.Abort()
 	if !(dm.nabort == 1 && txn.Status() == Aborted) {
 		t.Fatalf("abort: nabort=%d; txn.Status=%v", dm.nabort, txn.Status())