Fix for bad jets in 2017 data
While running 2017 data production, we found ~4 events that have jets with insanely high energy (~90 TeV). We noticed them because they also have a leading constituent with energy = inf, and therefore caused fastjet to crash.
This patch removes those jets from the collection, and reuses the PFCaloMETRatio branch to denote events where these jets were found. Eventually a more permanent solution will be employed (once these events are submitted to the MET scanners, etc.).