csv_to_sqlite should convert values according to schema

If the schema is detected or specified, the rows.utils.csv_to_sqlite function must use the field classes to convert the values before exporting to SQL.

To reproduce the bug, create two files:

echo -e "value\n123\n   \n456" > data.csv
echo -e "field_name,field_type\nvalue,decimal" > test-schema.csv

Then, call the function via CLI:

rows csv2sqlite --schemas=test-schema.csv data.csv data.sqlite
# Prints: "ValueError: could not convert string to float: "

Note: doing this conversion always will slow down the process, but is the more robust way of converting if a schema file is specified (not the case when schema is detect). May add a --convert-from-schema parameter to run this conversion.