arrow

ARROW-3811 [R]: Support inferring data.frame column as StructArray in array constructors

So that `array()` and `chunked_array()` handle data frame columns (struct arrays in arrow speech), and record batches and tables handle them too.

``` r
library(arrow, warn.conflicts = FALSE)
library(tibble)

df <- tibble(x = 1:10, y = letters[1:10])
a <- array(df)
a$type
#> arrow::StructType
#> struct<x: int32, y: string>
a$as_vector()
#>     x y
#> 1   1 a
#> 2   2 b
#> 3   3 c
#> 4   4 d
#> 5   5 e
#> 6   6 f
#> 7   7 g
#> 8   8 h
#> 9   9 i
#> 10 10 j

batch <- record_batch(a = rnorm(10), y = df)
batch$schema
#> arrow::Schema
#> a: double
#> y: struct<x: int32, y: string>
as_tibble(batch)
#> # A tibble: 10 x 2
#>         a   y$x $y
#>     <dbl> <int> <chr>
#>  1 -0.747     1 a
#>  2  0.285     2 b
#>  3  0.159     3 c
#>  4  0.707     4 d
#>  5  0.366     5 e
#>  6  0.237     6 f
#>  7 -0.730     7 g
#>  8 -0.880     8 h
#>  9 -0.833     9 i
#> 10  1.21     10 j

tab <- table(a = rnorm(10), y = df)
tab$schema
#> arrow::Schema
#> a: double
#> y: struct<x: int32, y: string>
as_tibble(tab)
#> # A tibble: 10 x 2
#>         a   y$x $y
#>     <dbl> <int> <chr>
#>  1  0.828     1 a
#>  2  1.57      2 b
#>  3 -1.54      3 c
#>  4  0.723     4 d
#>  5  1.07      5 e
#>  6  1.08      6 f
#>  7 -0.338     7 g
#>  8 -2.26      8 h
#>  9  0.198     9 i
#> 10 -1.58     10 j
```

<sup>Created on 2019-06-25 by the [reprex package](https://reprex.tidyverse.org) (v0.3.0.9000)</sup>

Author: Romain Francois <romain@rstudio.com>

Closes #4690 from romainfrancois/ARROW-3811/struct_column_inference and squashes the following commits:

b24645a2 <Romain Francois> + test using data frame column in record batch
773ffdee <Romain Francois> implement CheckCompatibleStruct()
6e54b209 <Romain Francois> array() handles data frame columns -> struct arrays
2d3bdd9a <Romain Francois> type() can infer struct column types from data frames