Bugfixes and improvements to summary_rows()

[rich-iannone]

This addresses a bug (#142) where aggregation function would operate on all columns and retain only the results for the requested columns. The prior behavior easily results in failures when an aggregation function operates on any column with incompatible data.

The PR also adds functionality for grand summaries (#38, #68). If groups == FALSE we obtain a collection of summary rows that operate on all of the data in the table. We can have both groupwise summaries and grand summaries in the same output table. Grand summaries are distinguishable from groupwise summaries by use of a large double border (and placement at the bottom of the table)

Finally, the PR applies several fixes for summary label generation and catches several edge cases. It's both more flexible, less prone to failure, and stops (with nice messages) if some of the stricter rules aren't followed. The documentation has been updated to reflect these changes.

Here's a table using a groupwise summary (across all groups):

sp500 %>%
  dplyr::filter(
    date >= "2015-01-05" &
      date <="2015-01-16"
  ) %>%
  dplyr::arrange(date) %>%
  dplyr::mutate(
    week = paste0(
      "W", strftime(date, format = "%V"))
  ) %>%
  dplyr::select(-adj_close, -volume) %>%
  gt(
    rowname_col = "date",
    groupname_col = "week"
  ) %>%
  summary_rows(
    groups = TRUE,
    columns = vars(open, high, low, close),
    fns = list(
      min = ~min(.),
      max = ~max(.),
      avg = ~mean(.)),
    formatter = fmt_number,
    use_seps = FALSE
  )

Here's a table using a grand summary:

sp500 %>%
  dplyr::filter(
    date >= "2015-01-05" &
      date <="2015-01-16"
  ) %>%
  dplyr::arrange(date) %>%
  dplyr::mutate(
    week = paste0(
      "W", strftime(date, format = "%V"))
  ) %>%
  dplyr::select(-adj_close, -volume) %>%
  gt(
    rowname_col = "date",
    groupname_col = "week"
  ) %>%
  summary_rows(
    groups = FALSE,
    columns = vars(open, high, low, close),
    fns = list(
      min = ~min(.),
      max = ~max(.),
      avg = ~mean(.)),
    formatter = fmt_number,
    use_seps = FALSE
  )

Fixes #38
Fixes #68
Fixes #142

[rich-iannone]

Some design considerations:

• columns should default to TRUE (this has the same effect as NULL: including all columns)
• groups as FALSE should no-op (NULL means grand summary, TRUE is all groups, c() specifies individual groups)
• a location helper function for grand summary rows is necessary as are CSS rules for the new content areas

@jcheng5 please feel free to add anything I missed or refine what's here.

[rich-iannone]

What if there are no groups but we call summary_rows() (and don't set groups = NULL)? I think that instead of stopping we should just provide the grand summary.