Flowserv-core

* Remove additional sub-directory `.flowserv` from default data directory path.
* New workflow template parameter type `actor` to support complete workflow steps as argument values.
* Serial engine workers to support execution of individual workflow steps using different backends.
* Remove separate Docker workflow controller

* Change the location of the default data directory to be the OS-specific user cache directory.

* Add option for passing an access token in API factory call.
* Ensure not to override *FLOWSERV_ASYNC* in `ClientAPI`.
* Add CLI environment context to support entry points for `flowserv` and `rob`.
* Extend serialized objects to contain additional resources (i.e., groups and runs) for authenticated users.

* Introduce `flowserv.service.api.APIFactory` to support multiple clients with different configuration on the same machine.
* Add functionality for user groups and authentication via the API client.
* Better support for accessing run result files via the API client.
* Include full functionality from [rob-client](https://github.com/scailfin/rob-client) into the `flowserv` CLI (\#72).
* Provide support for local and remote API's.
* Remove some of the obsolete code for running workflows in Jupyter Notebooks (now included in the new client interface `flowserv.client.Flowserv`).
* Drop support for Python 3.6.

* Add parameter declarations for lists and records (\73).
* Rename run argument serialization helper functions.

* Create directory for SQLite database files (\68).
* Type tests for enum and numeric parameters.
* Add serialized file handle to file listing in the application run result.
* Fix bug in run handle serialization for output files.
* Fix bug in Docker engine (return rundir in error case).
* Add ignore postproc option to `flowserv install` (\69).
* File object for files that are uploaded as part of Flask requests.
* Make database web_app flag configurable using environment variable (\70).