rewrite as_polars_series() for nested types

[eitsupi]

Related to #1012

Currently, due to the following processing, the types in list are processed ignoring the S3 method of as_polars_series, so appropriate type conversion is not performed.

r-polars/src/rust/src/conversion_r_to_s.rs

Lines 177 to 193 in 3a08926

	Rtype::List => {
	// Recusively handle elements of list
	let result_series_vec: pl::PolarsResult<Vec<SeriesTree>> = collect_hinted_result(
	x.len(),
	x.as_list()
	.unwrap()
	.iter()
	.map(|(name, robj)| recursive_robjname2series_tree(&robj, name)),
	);
	result_series_vec.map(|vst| {
	if vst.is_empty() {
	SeriesTree::SeriesEmptyVec // flag empty list() with this enum, to resolve polars type later
	} else {
	SeriesTree::SeriesVec(vst)
	}
	})
	}

Like this:

polars::as_polars_series(list(data.frame(a = 1), data.frame(a = 2)))
#> polars Series: shape: (2,)
#> Series: '' [list[list[f64]]]
#> [
#>  [[1.0]]
#>  [[2.0]]
#> ]
arrow::as_arrow_array(list(data.frame(a = 1), data.frame(a = 2)))
#> ListArray
#> <list<item: struct<a: double>>>
#> [
#>   -- is_valid: all not null
#>   -- child 0 type: double
#>     [
#>       1
#>     ],
#>   -- is_valid: all not null
#>   -- child 0 type: double
#>     [
#>       2
#>     ]
#> ]
polars::as_polars_series(arrow::as_arrow_array(list(data.frame(a = 1), data.frame(a = 2))))
#> polars Series: shape: (2,)
#> Series: '' [list[struct[1]]]
#> [
#>  [{1.0}]
#>  [{2.0}]
#> ]

^{Created on 2024-04-09 with reprex v2.1.0}

We probably need to use as_polars_series.list and a function that constructs Series from R list of RPolarsSeries in Rust side.