Make SDV compatible with SDMetrics 0.12.1

[npatki]

Problem Description

SDMetrics 0.12.0 hasn't been released yet (as of writing this issue). When it is released, we'll have to make some updates in the SDV visualizations wrappers to support the new functionality.

Note that SDV has a pinned requirement of sdmetrics<0.12. So we'll need to bump this up too when we're done.

Expected behavior

Update the visualization functions for both Single Table and Multi Table.

• get_column_plot:
  • Add an optional parameter called plot_type
    • By default, we should determine the plot type based on the metadata. (a) if sdtype is numerical/datetime, use 'distplot', (b) if sdtype is categorical/boolean, use 'bar', (c) otherwise, the data is incompatible so raise an Error.
    • If the user provides a plot type, then their provided value overrides the logic above
  • For datetime columns, SDV should convert the data to a datetime64 dtype, using the provided datetime_format from the metadata. Pass the converted data along to the SDMetrics visualization
• get_column_pair_plot
  • Similar to above, add an optional parameter called plot_type
    • By default, determine plot type based on meatdata. (a) if both sdtypes are numerical/datetime, then use 'scatter', (b) if both sdtypes are categorical/boolean, then use 'heatmap', (c) if one is numerical/datetime and the other is categorical/boolean, use 'box', (d) otherwise, raise an Error because the data is incompatible
    • If the user provides a plot type, t hen their provided value overrides the logic above
  • For datetime columns, SDV should convert the data to a datetime64 dtype, using the provided datetime_format from the metadata. Pass the converted data along to the SDMetrics visualization
• (multi-table only) get_cardinality_plot
  • Add an optional parameter called plot_type
    • By default, the plot type should be 'bar' in all cases
    • If the user provides a plot type, then pass that along instead

Additional Context

For the error messages, we may want to create a new type of error for visualizations.

>>> fig = get_column_plot(
    real_data=real_data,
    synthetic_data=synthetic_data,
    column_name='user_id',
    metadata=metadata,
)

VisualizationUnavailableError: The column 'user_id' has sdtype 'id', which does not have a supported visualization. To visualize this data anyways, please add a 'plot_type'.