implement anhecova robust variance and its report

[clarkliming]

this feature is dependent on #7, however, the work can be started based on some fake data (unbiased predictions)

• implement the robust variance
• the argument should at least contain "randomization_schema", "formula" (randomization formula, treatment ~ strata, or other forms that can clearly define these information
  • other input, like y, residuals and data are also important
  • not necessarily be a generic or method, because these values can be obtained relatively easy
• follow the logic defined in the RobinCar (and the paper)
• expose the function and add readme

[hoppanda]

• add reporting funciton
• add unit tests