When testing the Query Settings mechanism on time series collections, I have noticed that some queries cannot be fully answered by indexes, while the same query on a normal collection can. One example is the following $lookup command, where both collections are time series collections:

pipeline: [
      { $match: { a: 1, b: 5 } },
      {
        $lookup:
          { 
            from: secondaryColl.getName(),
            localField: "a",
            foreignField: "a",
            as: "output"
          }
      }
]; 

The $lookup stage results into sub-query, with the following Canonical Query:

[j0:prim] ns=test.system.buckets.secondaryTree: $or
[j0:prim]     $and
[j0:prim]         control.max.a $_internalExprGte 1.0
[j0:prim]         control.min.a $_internalExprLte 1.0
[j0:prim]     $expr {$ne: [{$type: ["$control.min.a"]}, {$type: ["$control.max.a"]}]}
[j0:prim] Sort: {}
[j0:prim] Proj: {}
[j0:prim] Collation: { locale: "simple" }

given the following query settings:

{ indexHints: [ { ns: { db: "test", coll: "secondary" }, allowedIndexes: [ { a: 1.0 } ] } ] } 

The interesting part is the the first branch of the $or condition can be fully answered using indexes, but the second can't, which seems to be added only for time series collections: $expr {$ne: [{$type: ["$control.min.a"]}, {$type: ["$control.max.a"]}]}

As this issue was blocking my ticket SERVER-86996, I've left some comments which should be fixed by this ticket.

These issue occur when running the `jstests/core/query/query_settings/query_settings_index_application_aggregate.js` test with the `timeseries_crud_jscore_passthrough` suite.