Sadedegel

* **FIX**: Fix a severe bug causing sadedegel summmarizers not to preserve sentence order in `sadedegel.server`calls.
* **ADD**: Add Community Contributors section to reference
* **ADD**: Freeze dependency versions in `requirements.txt`
* **UPDATE**: If SBD couldn't find a sentence assume whole document as a sentences.
* **UPDATE**: Reduce precision in summarizer evaluation table.

* **ADD**: Evaluation results of yet another baseline `LengthSummarizer` 👍 👍 👍
* **ADD**: Several less critical pending pull requests
* **FIX**: Documentation errors

While we are adding lots of things we significantly massed up
1. Code organization
2. Unit test coverage (drop down to **69%** 👎 )

A sequence of maintenance releases on top of base minor release 0.10 helped us to climb back in coverage and tidy up (Current coverage is above **90%** 👍 )

A few more things

* **ADD**: [A Notebook Explanining Basic Summarization with sadedeGel](notebook/Summarizers.ipynb)
* **CHANGE**: We introduced/relocate two new modules called `sadedegel.bblock` (sadedeGel Building Blocks) and `sadedegel.ml` and move all basic things into `sadedegel.bblock`. `sadedegel.ml` turns out to be a requirement to prevent cyclic reference.

Another big release
* **ADD**: 3 summarization algorithm based on clustering of BERT embeddings
* **ADD**: Related capabilities are added to `Doc` and `Sent` classes to support those algorithms.

* **ADD**: `info` subcommand for `sadedegel` command. This returns currently installed version, highest available version and web service version on [Heroku](https://sadedegel.herokuapp.com)
* **ADD**: Introduction of [Summarizer Performance Evaluation](sadedegel/summarize/README.md)
* **ADD**: `sadedegel.summarize evaluate` is introduced to regenerate results we provide.
* **UPDATE**: Summarizer abstract class is introduced.

Major feature introduced with version 0.7 is the GA of human annotated summary corpus in `sadedegel.dataset`
* **ADD**: Annotated json file corpus for validating various summarizers
* **ADD**: Corpus validation for annotated corpus
* **UPDATE**: `file_paths` in `sadedegel.dataset` now works for all corpus types.

Other changes in this release
* **FIX**: Proper behaviour for `rouge1_score` in edge cases.
* **ADD**: Github Action on master branch to auto push code coverage results to [codecov](https://codecov.io/gh/globalmaksimum/sadedegel/tree/master/sadedegel)
* **ADD**: Binder integration & Badge on repository [Notebooks](notebook/)
* **ADD**: Citation to open source supporting vendors we utilize in sadedeGel