[Proposal for Feature Addition]: Stratification by lab measurement thresholds

[mdlavallee92]

Cohort subsetting has been very useful when working on a study using circe cohorts. One feature that is not currently implemented (but comes up a lot) is stratifying by a measurement value. For example, stratify a target cohort by BMI thresholds [underweight (<18.5), normal (18.5-24.9), overweight (25.0-29.9) and obese (>=30)].

Requirements:

• target cohorts to stratify
• measurement concept ids
• timing of measure (i.e. on or closest to index is probably the default)
• unit concept ids (only include valid units)
• option for plausible value bounds (min =0, max = 50)
• strata categories

Questions:

• should a missing category also be made as a cohort?
• what other time bound option be considered?
  - first in window [-365 to -1; for example]
  - last in window

Will try to put together a PR based on some general code we have used in a project. Might need some help with the R6 classes.

Tagging @katy-sadowski and @alondhe for awareness.