Fast-carpenter

Added
- An API for different "backends" which run the actual data processing
- Add a backed for Coffea with local-multiprocessing executor. PR 102 [BenKrikler](https://github.com/benkrikler)

Changed
- Refactor AlphaTwirl code to use the "backend" API. PR 101 [BenKrikler](https://github.com/benkrikler)

Changed
- Added support for variables with multiple dots in the name (nested branches). Issue 95, PR 97 [kreczko](https://github.com/kreczko)
- Fix JaggedNth to work with arbitrary depth jagged arrays, Issue 87, PR 99 [benkrikler](https://github.com/benkrikler)
- Add protection against multiple dimensions using the same output name in a BinnedDataframe stage, Issue 92, PR 100 [benkrikler](https://github.com/benkrikler)

Added
- Existing collectors may now return their results within python, in addition to writing them to disk PR 90 [lgray](https://github.com/lgray).
- This behavior can, at present, only be controlled within python, and is meant for exposing certain aspects of FAST-carpenter plumbing to [coffea](https://github.com/CoffeaTeam/coffea).

Changed
- Fix bug in BinnedDataframe stage, issue 89, PR 93 [benkrikler](https://github.com/benkrikler)
- Pin atuproot to v0.1.13, PR 91
- Tidy the print out at the end of processing, PR 94.

Added
- Give fast-flow the "fast_carpenter" backend, issue 84, PR 85 [lorenafreitas](https://github.com/lorenafreitas)

Added
- JaggedNth supports negative indexing, PR 81 [pmk21](https://github.com/pmk21!)

Added
- Added version flag to CLI, PR 79. [maikefischer](github.com/maikefischer)
- Prohibit ncores < 1, PR 76 [annakau](https://github.com/annakau)

Changed
- Binned dataframes can now be produced from ND jagged arrays