Option to support extensions of `dplyr` verbs in `galah`

[mjwestgate]

At a recent webinar, @huizezhang-sherry suggested that we could implement extensions dplyr functions pretty easily, given that our functions use the same methods and terminology anyway. E.g. instead of:

galah_call() |> 
  galah_filter(year >= 2020) |>
  atlas_counts()

We could then use:

galah_call() |> 
  filter(year >= 2020) |>
  atlas_counts()

The approach is documented here.

Some other thoughts:

• select/galah_select, group_by/galah_group_by obvious extensions
• Other functions may not have a dplyr equivalent but could have alternate names, e.g. identify/galah_identify (also an S3 function from graphics) or apply_profile/galah_apply_profile
• Some things that are currently arguments could be replaced by functions, e.g. atlas_counts(expand = TRUE) could be replaced by expand() |> atlas_counts() (currently an S3 method in tidyr)

[mjwestgate]

Note to self to support slice_head() to atlas_occurrences() using pageSize arg in new API. Using the start arg it might be possible to support slice() as well.

On a related note; technically we've implemented slice_max() so far, despite calling it slice_head(), because solr ranks in decreasing order of count by default. To choose between these would could use fsort=count for slice_max() or fsort=index for slice_head(). Optionally we could add dir=asc to implement slice_min() & slice_tail()

[mjwestgate]

complete