[R] `col_types` ignored when `convert_options` specified

### Describe the bug, including details regarding any error messages, version, and platform.

Hey there,

I recently found out that my column specification won't work if I specify comma as the decimal mark—don't know if the same issue arises with other `convert_options` tho.

Reprex
````
# using write_csv2 to reproduce data with comma as decimal and semicolon as separator
readr::write_csv2(
tibble(
a = c("001", "100", "020", "3"),
b = c("ana", "maria", "joão", "açafrão"),
c = c(123.45, 122.80, 121.5, 6)
),
"data-raw/test.csv"
)

# won't work
df <- open_delim_dataset(
"data-raw/test.csv",
delim = ";",
col_types = schema(a = string()),
convert_options = arrow::csv_convert_options(decimal_point = ",")
)

schema(df)

# works
df <- open_delim_dataset(
"data-raw/test.csv",
delim = ";",
col_types = schema(a = string()),
)

schema(df)
````

### Component(s)

R