Ploomber

- `File` now supports a `client` argument to upload products to cloud
storage
- Adds `GCloudStorageClient`
- Fixes error that caused jupyter to fail to initialize the dag when
adding a function to a module already included in the YAML spec
- Fixes IPython namespace errors when using `ploomber interact`
- Adds `ploomber.testing.sql.assert_no_duplicates_in_column` to check
for record duplicates and optionally show duplicates statistics
- Deprecates a few internal methods: `Table.save`, `DAG.to_dict()`,
`Task.to_dict()`
- Improvements to SQL static analyzer to warn when relations created
by a SQL script do not match `Product`
- A few changes to `Metadata` (internal API) to cover some edge cases
- Warning when `Product` metadata is corrupted
- Adds new `meta.import_tasks_from` option in YAML specs to import
tasks from another file

- Deprecates `ploomber new` and `ploomber add`
- Adds `ploomber scaffold`
- Jupyter plugin now exports functions as notebooks using
`jupyter_functions_as_notebooks` in `pipeline.yaml`

- `ploomber add` generates template tasks and functions if they don't exist
- Jupyter plugin now shows PythonCallable tasks as notebooks

- Documentation tutorials re-organization and CSS fixes
- Improvements to the `InMemoryDAG` API
- Minor bug fixes
- `File.__repr__` shows a relative path whenever possible

- Adds support for passing glob-like patterns in `ploomber build` (via
`DAGSpec.from_directory`)

- Full Windows compatibility
- Adds documentation to show how to customize notebook output using
`nbconvert`
- Improvements to introductory tutorials
- Adds `--debug/-d` option to `ploomber build` to drop a debugger if
an exception happens
- Ensuring all dag-level, task-level and product-level clients are
closed after `dag.build()` is done
- Minor bug fixes