Reduce load on SPs

[bajtos]

As the Station network grows, we put a lot of load on SPs. Some CIDs are being retrieved 1000x a day, which is rather wasteful. SPARK is about retrieval testing not load testing.

Let's improve this.

• Add per-round configuration "tasks per node" (TN) as described in SPARK tasking v2.
• Modify spark checkers to honour this config when calculating delay between retrievals
• Modify spark checkers to fetch to root block only
• Deploy SPARK to Stations
  • publish a new SPARK version
  • publish a new Station Core version
  • publish a new Station Desktop version
• Reward only first TN measurements per round filecoin-station/spark-evaluate#98
• Add a new metric to track how many measurements are submitted for each task + visualise it
• feat: reject measurements from spark versions <1.7.0 filecoin-station/spark-api#173

[bajtos]

https://github.com/filecoin-station/core/releases/tag/v16.3.0