use `vroom::col_big_integer()` for integers to avoid overflow?

[khusmann]

I notice in #16 you switched from col_integer() to col_double() to avoid integer overflow.

readr doesn't support big integers, but vroom does via vroom::col_big_integer(): https://vroom.r-lib.org/articles/vroom.html#column-types

We may want to try out col_big_integer() out instead of col_double(); it appears to work with the gps data you posted in #16, at least.

[peterdesmet]

1. Though undocumented, vroom::col_big_integer() can be passed as a collector to read_csv() 👍

2. The switch from col_double() to vroom::col_big_integer() would be here:

   frictionless-r/R/collectors.R

   Lines 44 to 46 in b7fce08

   	} else if (bare_number) {
   	readr::col_double() # Not col_integer() to avoid big integers issues
   	} else {

3. Neither vroom::col_big_integer() nor col_integer() currently allow values with a positive sign (+1 or +01). As a result, this test fails for the first value in the second row (+03):

   frictionless-r/tests/testthat/data/type_integer.csv

   Lines 1 to 3 in b7fce08

   	int,int_neg,int_factor,int_ws,int_notbare
   	3,-3,3, 3 ,$3
   	+03,-03,4,3 ,3 dollar

   This has been reported col_integer() does not allow positive sign (+) while parse_integer() does tidyverse/readr#1587