ARROW-5505: [R] Stop masking base R functions

[nealrichardson]

New attempt at this. It replaces table() and array() with to_arrow(), which already existed but was not exported.

Review much appreciated @romainfrancois

[codecov-io]

Codecov Report

Merging #5172 into master will decrease coverage by 12.43%.
The diff coverage is 100%.

@@ Coverage Diff @@## master #5172 +/- ##
==========================================
- Coverage 87.63% 75.2% -12.44% 
==========================================
Files 1022 58 -964 Lines 146258 3621 -142637 Branches 1437 0 -1437 ==========================================
- Hits 128180 2723 -125457 + Misses 17716 898 -16818 + Partials 362 0 -362

Impacted Files	Coverage Δ
r/R/Table.R	85% <ø> (-3%)	⬇️
r/R/write_arrow.R	100% <ø> (+3.44%)	⬆️
r/R/array.R	62.5% <ø> (-1.14%)	⬇️
r/R/to-arrow.R	100% <100%> (ø)
r/src/table.cpp	75.75% <0%> (-5.06%)	⬇️
python/pyarrow/ipc.pxi
cpp/src/parquet/column_page.h
cpp/src/plasma/test/external_store_tests.cc
cpp/src/arrow/array/builder_decimal.cc
cpp/src/plasma/client.cc
... and 961 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c050244...209f182. Read the comment docs.

[romainfrancois]

I get the need to not masking base functions and this is something we have come to regret in other projects. I was trying here to mimic the 🐍 interface (pa.array() ...)

to_arrow() feels vague, maybe we could do some sort of namespacing, e.g. arrow_array() or something, but then we would kind of have to do it for all functions :/

[wesm]

This has a rebase conflict, what's the status of this?

[nealrichardson]

Closing; this is superseded by #5279.