Pyclowder

Fixed

- Extractor would not run in case clowder_url was ""

Fixed

- Thread of heartbeat was set as daemon after start.

Fixed

- When rabbitmq restarts the extractor would not stop and restart, resulting
in the extractor no longer receiving any messages. 17

Added

- Can specify url to use for extractor downloads, this is helpful for instances
that have access to the internal URL for clowder, for example in docker/kubernetes.

Removed

- Removed ability to run multiple connectors in the same python process. If
parallelism is needed, use multiple processes (or containers).

With this version we no longer gurantee support for versions of python below 3.

Fixed

- There was an issue where status messages could cause an exception. This would
prevent most extractors from running correctly.

**CRITICAL BUG IN THIS VERSION. PLEASE USE VERSION 2.3.1**

Removed develop branch, all pull requests will need to be against master from now
forward. Please update version number in setup.py in each PR.

From this version no more docker images are build, please use pip install to
install pyclowder.

Added

- Simple extractors now support datasets, can also create new datasets.
- Ability to add tags from simple extractor to files and datasets.
- Ability to add additional files (outputs) to dataset in simple extractor.
- Use pipenv to manage dependencies.
- Add job_id to each status message returned by pyclowder.
- PyClowderExtractionAbort to indicate the message shoudl not be retried.

Changed

- Better handling of status messages

Fixed

- Heartbeat of 5 minutes would cause timeouts for RabbitMQ
[CATSPYC-30](https://opensource.ncsa.illinois.edu/jira/browse/CATSPYC-30)
- support uploading large file to dataset.
[CATSPYC-29](https://opensource.ncsa.illinois.edu/jira/browse/CATSPYC-29)