Do not continue to delegate after finalize initiation

[peterbroadhurst]

Fixes #518

Inspecting this code, it seems like the intent of finalize() (linked below) is to:

1. Initiate a background finalization
2. Fire an event on success/failure of the finalization

However, it did not seem to be setting the finalizePending flag, or exiting the state reconcile loop.

paladin/core/go/internal/privatetxnmgr/transaction_flow_actions.go

Lines 250 to 287 in bf09215

	func (tf *transactionFlow) finalize(ctx context.Context) {
	if tf.finalizeRevertReason == "" {
	log.L(ctx).Debugf("finalize transaction %s", tf.transaction.ID.String())
	} else {
	log.L(ctx).Errorf("finalize transaction %s: %s", tf.transaction.ID.String(), tf.finalizeRevertReason)
	}
	//flush that to the txmgr database
	// so that the user can see that it is reverted and so that we stop retrying to assemble and endorse it
	tf.syncPoints.QueueTransactionFinalize(
	ctx,
	tf.transaction.Domain,
	tf.domainAPI.Address(),
	tf.transaction.ID,
	tf.finalizeRevertReason,
	func(ctx context.Context) {
	//we are not on the main event loop thread so can't update in memory state here.
	// need to go back into the event loop
	log.L(ctx).Infof("Transaction %s finalize committed", tf.transaction.ID.String())
	// Remove this transaction from our domain context on success - all changes are flushed to DB at this point
	tf.domainContext.ResetTransactions(tf.transaction.ID)
	go tf.publisher.PublishTransactionFinalizedEvent(ctx, tf.transaction.ID.String())
	},
	func(ctx context.Context, rollbackErr error) {
	//we are not on the main event loop thread so can't update in memory state here.
	// need to go back into the event loop
	log.L(ctx).Errorf("Transaction %s finalize rolled back: %s", tf.transaction.ID.String(), rollbackErr)
	// Reset the whole domain context on failure
	tf.domainContext.Reset()
	go tf.publisher.PublishTransactionFinalizeError(ctx, tf.transaction.ID.String(), tf.finalizeRevertReason, rollbackErr)
	},
	)
	}

[hosie]

LGTM