Creating this issue now so I can note the following: we'll want .align_and_extend_estfuns() to order units of observation by unit of assignment so that clustered HC2 computations are more straightforward (we can create then concatenate $(I - P)_{i}^{-1/2}$ matrices for each unit of assignment $i$ and multiply them directly with the output of .align_and_extend_estfuns()). This means that if the by argument is specified in cov_adj(), we'll want to use it to align contributions to estimating equations but not order them.
Creating this issue now so I can note the following: we'll want
.align_and_extend_estfuns()to order units of observation by unit of assignment so that clustered HC2 computations are more straightforward (we can create then concatenate $(I - P)_{i}^{-1/2}$ matrices for each unit of assignment $i$ and multiply them directly with the output of.align_and_extend_estfuns()). This means that if thebyargument is specified incov_adj(), we'll want to use it to align contributions to estimating equations but not order them.