This should be a little more accessible an entry point for client side stuff then staging Galaxy jobs as an example.

Includes a generalizations for how dynamic outputs are collected from the working directory, this was going to be needed to implement discovered datasets on the Galaxy side anyway.

Not a hugely important test but it does ensure the modules load cleanly and that the scripts all respond to --help with a 0 exit code.

Fixes #62.

Rev verison, tweak docs, add documentation describing how to release.

 - Add a new section to the docs for configuring Pulsar, remove this from README.
 - Replace full install information in README with a very short quickstart and link to new install section of docs.
 - Add app.yml.sample - document parameters there instead.
 - Remove non-web related configuration from server.ini.sample.
 - Replace description in job_managers.rst with app.yml based description.
 - Preserve older variant of job_managers documentation as docs/job_managers_old.rst.
 - Add script to automatically generate RST documentation from scripts.
 - Add auto-generated documentation for various pulsar scripts (pulsar-main, pulsar-check, pulsar-config).
 - Fill out detailed option help for pulsar-check, pulsar-main, and pulsar-config.
 - Add manually created documentation for the puslar script.
 - Update pulsar-config to only present options available on current platform (e.g. no supervisor on Windows)
 - Many more tests for pulsar-config, comment out unreachable code.

Closes #68.
Closes #67.
Closes #23.

Will help ensure these jobs eventually become terminal.

Start process of describing PyPI based installs and newer scripts such
as pulsar-config and pulsar-check.

 - Fix bug where jobs may have been recovered before the MQ callback was set on the manager.
 - Add a new LOST job state for jobs that cannot be recovered and the Pulsar is going to abandon.
 - Fire off callback for LOST jobs.
 - Add test case to verify callback is called for LOST jobs.

By default pulsar will use 0777 with current umask filtered out.

Closes #65.

It is disappointing it has been so broken for so long, but at least with this fix it sounds like it is working for @dctrud.

Fixes #63.

Should hopefully address the crazy issue #59.

Probably not needed - only hit the error condition because of other bugs - but I guess it is not hurting anything.

Forces a full setup step without having to actually create a non-empty setup_params dictionary. Galaxy would always pass a tool id and stuff in setup_params - but these should be optional.

Want to reuse it in a test.

Useful for debugging deadlock situtations.

(In case I need it in the future again.)

Eliminates required dependency on paste for Python 3 (though still need to update actual code for Python 3).