File-based CSV: Strings should not be null'd by default

[girarda]

What

The CSV parser converts values to None if they match any of the values in the null_values set, which is configurable as part of the CsvFormat.

To be backward compatible with the legacy S3 source, string values should be kept as-is by default, even if they match one of the values in the null_values set.

We should add an option to the CsvFormat to make the strings nullable

Acceptance Criteria

• By default, strings are not turned to None even if they match a null_values
• The CsvFormat has a boolean field "strings_can_be_null". When True, strings matching a null_value are None

[girarda]

grooming notes:

• by default, we will set strings_can_be_none to True
• legacy default: set strings_can_be_none to False

[girarda]

Update: we'll call the field strings_can_be_null instead of strings_can_be_none