api: Add fixed # tasks per round. Closes space-meridian/roadmap#82

[juliangruber]

Closes space-meridian/roadmap#82 Depends on https://github.com/filecoin-station/spark-stats/issues/214

Alternatives considered:

• Base task count on measurement count from previous round. This data doesn't exist in spark-stats yet. It should lead to better results though, than the currently used station-id count. @bajtos wdyt? Shall we give this a try and see if it behaves well enough? We went with this alternative

Questions:

• Do we need to test this code path? Yes

TODO

• [x] Track in Graphana
  • [x] measurements target for round
  • [x] measurements count for round
  • [x] total desired measurement count
  • [x] actual measurement count
• [x] Write tests for scaling

[juliangruber]

+   maxTasksPerNode: 2,
-   maxTasksPerNode: 15,

Using live data, this will decrease the job count by a lot. We might want to update the target job volume

[juliangruber]

Thanks! Will ship tomorrow