Runtime upgrade E2E test should wait for old node expiration

[kostko]

Currently the runtime upgrade E2E test shuts down the old compute nodes after upgrade and starts the client, assuming that the new compute nodes will process the transactions:

oasis-core/go/oasis-test-runner/scenario/e2e/runtime/runtime_upgrade.go

Lines 211 to 220 in e79b497

	// Kill old compute workers.
	sc.Logger.Info("stopping old runtimes")
	for i := 0; i < sc.firstNewWorker; i++ {
	worker := sc.Net.ComputeWorkers()[i]
	if err = worker.Stop(); err != nil {
	return fmt.Errorf("stopping old compute worker: %w", err)
	}
	}
	// Start the new compute workers.

The problem is that the committee scheduler can still schedule one of the old nodes, making the test flaky as it depends on what committee is elected. We should probably wait for the old nodes to expire before proceeding.

[Yawning]

This blocks #3180