For a stellar code, if a node is configured with multiple history archives for reading, presumably from sources of different nodes. How does this node handle discrepancies between the historical data? In other words, if an archiving node is malicious and publishes falsified data. What are the measurements a good behaving node takes to validate this? Namely, does the good node download archived data from all of the archives in its configuration and compare them? And if not (or assume it has only one source of archive), how does it go about validating the data downloaded?