[STORY] Implement Dataset.apply() and arcana apply analysis

[tclose]

Description

API and CLI interfaces for applying Analysis classes to datasets need to implemented as described at https://arcana.readthedocs.io/en/develop/processing.html#analysis-classes

Acceptance Criteria

• 1. Analysis classes can be applied to datasets via API
• 2. Analysis classes can be applied to datasets via CLI
• 3. Newly added methods are covered to 90%

Dependencies

• [STORY] Create analysis class decorators #9
• [STORY] Clean up spefications of analysis classes #130

Tasks

• dataset apply method takes an analysis class and the args to be passed to it, then creates the object and saves it in the dataset
• add pipeline.stack method with column salience filter
  • builds stack of Pipeline objects with workflows created by builder methods
  • workflows merged into single pipelines where the outputs of the builder pipelines are below the Salience threshold and there is no merging over columns
  • Switches will need to be stored as part of Pipeline object
  • ColumnSpec objects to be returned alongside each pipeline object, signifying new columns to be added to the dataset
  • have method to display this nicely by passing the right flag to stack
• Save pipelines in dataset alongside independent pipelines. Use them be default when creating new Derivatives, unless a "rebuild_workflow" flag is passed or similar. Saved pipelines can be compared with those that would be generated by analysis class as quick way of picking up changes that need to e overwritten.
• add 'menu' method to analysis classes, make it dynamically determine which columns can be created and filter to these, unless the --all flag or similar has been set.
• "checks" method in columnspec to return all checks on column
• manual checks to be specified by basic check that simply returns the "questionable" value
• Support for 'subanalysis.column' in AnalysisSpec.column(). Return mapped ColumnSpec with '.' In name and "mapped_from" set to appropriate value
• map pipeline outputs when importing pipelines from sub-analyses be recreating a modified copy of the PipelineBuilder spec

[tclose]

Epic: Australian-Imaging-Service/pipelines#17