Ressurect Conflux Proposal

@dgoulet wrote an initial conflux proposal draft. In the network-team meeting on Nov 17th, we discussed reviving it, but based on Proposal 324.

Things we definitely need:

• Unreliable conflux (no resumption windows)
• Use Prop324 congestion window information to decide which circ to send data on
• Add sequence numbers to relay cells on each leg (Prop 325? Or other approach?)
• Minimize the reorder window at Exits for stream assembly (pushback tradeoffs, etc)
• Side channel analysis section

Nice to have:

• Pre-emptive circuit discussion
• Minimize client reorder window
• Resumption if a circuit leg dies

This is Sponsor 61, Objective 3.1 material.