Datalad-crawler

- Making compatible with recent DataLad by using new WitlessRunner and
not older unused features.

- RF: stop using `_{git,annex}_custom_command` to allow DataLad core
progress forward without "breaking" the crawler

- ENH: fix enabling special remotes when working ith recent (as of 202006)
git-annex
- NF: gh (for github) and xnat crawler pipelines

- DataLad 0.12 is now minimal version. Codebase is now compatible with current
DataLad 0.12.2-293-gd5fcb4833
- uses less of GitPython functionality
- OpenfMRI pipeline tests "relaxed" (no commit counts etc)
- s3 node - be robust in case of no previous version-id known

- ENH: `s3_simple` pipeline got additional option `drop_immediately` to
drop files immediately upon having them annexed
- RF: `mock` is explicitly listed as a dependency for testing since DataLad
0.12.x will be PY3 only and could use built-in `unittest.mock`

- MNT: More changes for compatibility with developmental DataLad (62)

0.4.2 (Oct 30, 2019) -- Friends should stick together ...

- BF: Prevent sorting error on missing attribute (45)
- BF: enclose "if else" into () since it has lower precedence than + (43)
- MNT: Adjust imports for compatibility with developmental DataLad (53)
- MNT: Update save() call for compatibility with new save (42)