Refactor epi[x]_slide to reduce code duplication

[brookslogan]

E.g.,

• Create utilities to process f, ..., & any other appropriate args into uniform types (e.g., process function&args/formula&args/missing&tidyeval into a uniform function f and ... args), apply in both epi_slide and epix_slide, and simplify away related branching so that group_modify & {slide,comp}_one_grp are referenced only once per slide function.
• Consider making some element/row-recycling utilities, although note differences in how this work in epi_slide vs. epix_slide (discussion in group_by() for epi_archive objects #64 and docs in associated feature branch may be helpful).
• Related: Allow multiple column computations in data masking slides #235.

[brookslogan]

A lot of this has been accomplished in handling #326, and other parts like row-recycling probably don't apply anymore, as we don't do that for epi_slide_opt, epix_slide.