Do not call networkstatus_reset_download_failures() hourly
Once an hour, we call router_reset_descriptor_download_failures, which calls networkstatus_reset_download_failures(). This probably isn't right! Our download logic has a way to say "Try this once an hour": that's by having a download schedule with a one-hour interval. What this does is make us retry several times an hour, as if we had never failed before.
Found while investigating legacy/trac#4580 (moved) .