Improve consensus timings

[pinebit]

🎯 Problem to be solved

To prevent missing proposals we shall revisit the timings used by consensus layer. Once the final values confirmed, we need to introduce a new timer type and a new version of QBFT protocol (see Consensus Abstraction design) that is using this new timer.

🛠️ Proposed solution

• [ ] Review the statistics on timings reported by Data.
• [ ] Implement new timer type and configure the new QBFT protocol version, say qbft/v2.1.0.
• [ ] Possibly change --consensus-protocol flag behavior to allow fully qualified protocol strings (so we can override with the precise qbft version).

🧪 Tests

• [ ] Tested by new automated unit/integration/smoke tests
• [ ] Manually tested on core team/canary/test clusters
• [ ] Manually tested on local compose simnet

[KaloyanTanev]

If we get this PR merged in v1.3.0, I believe we will need data from at least a month from couple of clusters, in order be able to advance with this request. That said, v1.5.0 seems like a reasonable milestone for this one.