The purpose of the product_limits test is to run queries that are very stressful to the server.

As Max rightfully observed, the WorkloadAndOverSingleField uses a predicate that is simplified to a trivial index range in the presence of an index:

db.foo.aggregate([{$match: {$and: [{i: {$lt: 1000}}, {i: {$lt: 1001}}, {i: {$lt: 1002}}]}}]).explain();
{
      inputStage: {
        ...
        indexBounds: { i: [ '[-inf, 1000)' ] }
      }
}
 

A single `indexBounds` is created and the predicate in its entirety is never evaluated anywhere within the query, making the test less stressful than it could be.

Contrast with the same workload run against a collection with no index:

       filter: {
        '$and': [
          { i: { '$lt': 1 } },
          { i: { '$lt': 2 } },
          { i: { '$lt': 3 } }
        ]
      },

The predicate is preserved and will be evaluated in its entirety.

So we need some predicate and/or dataset that is also stressful in the indexed case.