Taxonomy-resolver

- Re-added the "filter" option to the `TaxonResolver` class and CLI
- Several improvements in usability and consistency

- Improved performance with Nested Sets
- Added new `pandas` dependency
- Tree "schema" stored as a pandas ``DataFrame``
- Modified Preorder Tree Traversal markup is used for quick tree search
- Improved search times by reducing the number of subtrees searched
- Additional optimisations by using Python ``sets``
- Updated description in the ``README.rst``

- Removed the "Any Python Tree" `anytree` dependency
- Large refactoring and rewrite of the `TaxonResolver` class
- Removed `TaxonResolverFast` and `fast` mode
- New test cases based on a mock tree `testdata/nodes_mock.dmp`
- Simplified usage of the class and CLI
- Only currently working with `pickle` format for writing and loading
- Searching takes included, excluded and filter lists of TaxIDs
- Validation only takes included TaxIDs
- CLI `search` and `validate` can take multiple TaxIDs at once

- Additional fixes and improvements for missing and invalid TaxIDs

- Brought TaxonResolverFast in line with the functionality of TaxonResolver
- Searching and validation based on lists of TaxIDs (instead of relying only on files)
- Updated CLI to take multiple filter files
- Updated CLI to take comma-separated values for search and validation
- CLI writes to STDOUT by default for search and validation

- Capturing a list TaxIDs used for `filter`, which allows searching and validation without having to provide them again

- Fixed the logic under validation and searching
- Added `TaxonResolverFast` class for faster searching and validation (but it requires a built Tree)
- Added `--mode 'fast'` mode to the CLI