Consider slide_by(), .slide_by = / similar + mutate, summarize, reframe

[brookslogan]

An idea from {transmute,mutate}_scattered_ts was to have something like a slide_by() function followed by a normal {transmute,mutate}. We could apply that same idea to epi_slide and maybe epix_slide, allowing for {mutate,summarize,reframe,group_modify} in the {epi_df,epi_archive} methods for these, plus mirror dplyr 1.1 and have a .slide_by= parameter in each of these. At least sometimes we want a epi_slide_mutate and others an epi_slide_summarize, and this might be one way to organize this. But this interacts with other interface&naming issues that have not been settled. (E.g., should we have version_slide_by, time_slide_by, version_time_slide_by? Is there actually a mutate parallel for epix_slide? ...)

[brookslogan]

This could get even more complicated considering the non-reordering of .by.