chore(dynamic-sampling): reduce size of query sets for dynamic sampling

[shellmayr]

• Use 80 instead of 100 orgs, and 4000 projects intead of 5000 projects per query
• Intent is to reduce the size of the query, so snuba does not run into query timeouts as frequently when running these queries
  Closes TET-1224

[linear]

TET-1224 Make dynamic sampling work on lower org counts per task to prevent timeouts

[vgrozdanic]

You think that 20% reduction will be enough?

Do we want to go more aggressive and do a binary search? E.g. start with half and take it from there up or down

[shellmayr]

@vgrozdanic I get the idea, and I somewhat agree that we should do this if this were in a vacuum, but I want to be careful w.r.t. the cycles of the tasks. Each of these queries adds some time to the outer loop, so we should be careful not to disrupt the rhythms too strongly