Fix StandaloneFrontendHistory accounting
testTransactionForwardedToLeaderAfterRetry() is exposing a scheduling
difference between the two implementations -- with tell-based protocol
we are holding off sending modifications until there is a significant
event needing them to be flushed to the backend.
Fixing this up exposes a bug in StandaloneFrontendHistory transaction
accounting, as circles back to ShardDataTree, where its transactions
are accounted for again -- fix that mistake by allocating transactions
directly in StandaloneFrontendHistory.
JIRA: CONTROLLER-2017
Change-Id: Ied00910f55bd3c63f99ae187396c1a193e5ece20
Signed-off-by: Šimon Ukuš <simon.ukus@pantheon.tech>
Signed-off-by: Robert Varga <robert.varga@pantheon.tech>