Implement aggregate scaler and apply to q1 as an example

[mIXs222]

Implemented aggregate scaler which has 3 layers, from bottom to top:

• Cardinality estimator: take current group count and cardinality/progress fraction --> estimate the final group count
  • 3 flavors: constant, linear, and power (things in between constant and linear)
• Aggregate scaler: currently only supports SUM and COUNT scaling
  • This implements StreamProcessor<DataFrame> to pull progress fraction info
• Execution node: a wrapper

Added to q1.rs as an example. Verified that the mean absolute percentage error range is below 2%, instead of 60% without the scaler.

Depending on PR #133 (and PR #134)