API review: group sheet view settings, tidy docs and argument order

[billdenney]

An API review before Phase 6 widens things further. All pre-release, so no user
code breaks.

1. xl_sheet(): 23 arguments → 15

Phase 4's nine view settings went in flat, while page = xl_page_setup() had
already established the pattern for exactly this kind of cluster. They now
follow it: view = xl_sheet_view(...).

freeze deliberately stays on xl_sheet() — it is the single most common
worksheet option, and making the common case wordier to tidy the rare ones would
be a poor trade. split, its rarely-used divider variant, moves; the two stay
mutually exclusive and the error now names both spellings.

Done now so Phase 6's validation = lands in a function that isn't already
overloaded.

2. A stale documentation claim

write_xlsx(format_headers) still said the header would be "centered and bold".
That stopped being true when the header format became configurable via
xl_properties(header_format = ). It now describes what the argument does and
points at the setting; xl_workbook() @inheritParams it rather than carrying
its own paraphrase.

3. One concept, three argument names

xl_hyperlink(name =), xl_hyperlink_cell(value =) and
xl_cell_general(value =) all mean the text displayed instead of a URL. The
names differ only because xl_hyperlink() mirrors Excel's HYPERLINK()
argument. name is released API so it stays; the docs now state the
equivalence.

4. Duplicated coercion

if (is.factor(x)) x <- as.character(x); stopifnot(is.character(x)) appeared
verbatim in all three cell shorthands. Now .as_character_arg(), which also
reports the argument name and the class it got instead of stopifnot()'s
expression dump.

5. Most-probable-first argument order

• xl_font() leads with bold, italic, color, size, name rather than
  name, size, color, bold, italic
• xl_comment() leads with author before format

Every call site names its arguments past the first, so nothing positional
depended on the old order — verified by grep before changing them.

Correction to the review

I initially flagged the format argument as documented seven times with
inconsistent wording. On inspection that is mostly legitimate variation, not
duplication: xl_comment() uses a fill-and-font subset, xl_rich_run() a
font-only subset, and the others a full format. Unifying them would have lost
the distinction. Only col_names / format_headers were genuinely the same
argument described twice, and those are now inherited.