Caper

* Bug fix for Cromwell server IP address. (`None` -> `localhost` by default)
* Update README

* Default `singularity_cachedir` at `~/.caper/singularity_cachedir`
* Updated README

* Caper defaults to use a file DB. You can still use MySQL DB.

* added subcommand `troubleshoot` which will show list of errors that occurred during runtime.
* can have local relative paths in input JSON

* Bug fixes
- `--out-dir` didn't work with slurm, sge, pbs backends
- `--singularity-cachedir` didn't work correctly if `SINGULARITY_CACHEDIR` env var is already defined.

* added `--no-build-singularity` to prevent building singularity image before running a pipeline.
- but this can lead to conflicts between tasks trying to build their own local singularity image on the same target location.

* bug fix for singularity
- added `--singularity-cachedir` to prevent repeatedly building image for every WDL task.
- automatically calculate `SINGULARITY_BINDPATH` based on files in input JSON file
- by finding common roots of those files

* CaperURI
- more robust method to catch HTTP_ERROR_CODE from cURL.