Move HLT jet configuration to use getStep

FYI @fpastore, @tbold so you see how the final result looks 😄

This change should complete the refactor of the jet slice for deferred sequence generation. Here, the following restructuring was done:

• The JetChainConfiguration now builds the JetDefDict, so it knows the final jet collection construction sequence -- but it does not track its preselection (we could do a dictionary update to add this, actually, were it deemed necessary). This necessitates an extra call to prepareDataDependencies so that the flags can be provided.
• The bare minimum of information passing is retained, e.g. cluster collections don't have to be forwarded as a string, but could be extracted from the jet definitions. The only place something had to be returned together with a step is for the calo preselection, to forward the collection to roiftf step and avoid duplicating the extraction of the presel jet definition.
• The recordable check is moved to JetChainConfiguration, so that it checks only the terminating jet collection for any chain (N.B. this necessitates that the roiftf collections, which are in some test and exotics chains, are added to the EDM -- I put them only in BS/ESD for now).
• Step names are forced to change -- would propose to hold off on this until we can take out the stepID from getStep, which should simplify the comparison a bit.
• Syntax updates needed to go to TLA, MET, and UTT. @maklein I am keeping this drafted for the moment, but one of us might have to deal with minor conflicts for your own work on MET CA deferral.
• More input definitions had to go into JetRecConfig, @delsart
• Need to fix Heavy Ions, as the JetRecoDataDeps does not yet support that signature, and it is now called for every jet chain, where I was working around this before.

Also likely interested: @cantel, @pakontax, @miochoa