v0.1.8

Release v0.1.8

**`ReadableStream` support in `CSVTransformer`**: You can now directly pass
a `ReadableStream` as the return value of a transformation function. This
functionality is useful when the transformation itself involves a stream
(e.g., the transformation makes a network request).

**`CSVNormalizer` improvements**:
- The `CSVNormalizer` constructor now accepts a `string` shorthand value
as an alternative to a full `CSVNormalizerHeader` object. The string value
will be used as the column name. For example, passing `['columnA']` is
equivalent to passing `[{ name: 'columnA' }]`. You can also mix `string` and
`CSVNormalizerHeader` objects in the same array.
- `CSVNormalizer.toFieldMap()` now implements the `.delete()` and `.clear()`
functions to remove fields from the underlying row.

**Bug fixes**:
- Fix regression where `null` and `undefined` were unable to be denormalized
through `CSVDenormalizer`. These values are now denormalized into an empty
string `''`, as expected.
- Fix bug where batched rows were deleted before the transformation function
could use the data.
-----BEGIN SSH SIGNATURE-----
U1NIU0lHAAAAAQAAADMAAAALc3NoLWVkMjU1MTkAAAAg4OaSdijsW6raSPK3llrsZIEdtk
SQaGaXESF8DcruCNgAAAADZ2l0AAAAAAAAAAZzaGE1MTIAAABTAAAAC3NzaC1lZDI1NTE5
AAAAQD15HE+UReUSS1lhhZl8ElP3+YQtF+trs03ljOAOfkGQ+cPRijAuc5Vfki+NAzZZSm
DeyZHY4nSHNxSwPK3Q4wk=
-----END SSH SIGNATURE-----