Proof-of-time propagation

For the Proof-of-Time specification, one of the questions is how exactly the proofs-of-time will reach farmers.
As we have previously discussed:

  • domain operators (executors) are expected to dedicate a CPU core to run PoT; however, they are not explicitly incentivized to do so

  • operators, when elected to produce a transaction bundle, should include their latest PoT in the bundle. This way, an intermediary PoT result is produced roughly every second and available for verification.

  • including a wrong PoT is a punishable offense

These points bring up the following questions.:

  1. Should proof-of-time be included in the bundle header or as a signed extrinsic within the bundle body?

  2. What should be the appropriate lag value between when PoT is produced and included in the bundle and when the update of global randomness takes place?

  3. Should omission of PoT in a bundle be a punishable offense as much as a wrong PoT?