Ipf

* New lmod_cache_file parameter to ExtModules workflow; when specified, only
the modules in the spiderT table from the cache file will be published, and
nothing in the hiddenT table will be published.
* new configuration parameter --lmod_cache_file
* fix for init script: su command needed a -c argument
* IPF documentation revised for clarity and completeness
* ipf_configure default for support contact now JSON blob not URL

* fileread.py reads and parses generic json files, sample etc/ipf/workflow/templates/{generic_print,generic_publish}.json
* docs/GENERIC_PUBLISHER.md documents how to configure the new generic publisher
* all aspects of interaction between IPF and Slurm are parameterizable so that changes in the schedulers do not require new IPF code
* Improved configuration, bootstrapping of position files for monitoring slurmctl.log
* Modules improvements: Do not publish modules with a don’t publish flag
* Refactored IPF configuration (command line parameters to (re)configure single specific workflows)
* IPF documentation revised for clarity and completeness
* Convenience script for keeping ca_certs.pem up to date with XSEDE trusted CAs
* Now can use python-amqp 2.4 and above

* Bug fix for extmodules workflow where IPF was interpreting the Application
Name to be "ApplicationName/ApplicationVersion"

* Bug fix for how IPF handles an edge case that can occur in lmod_cache files (spiderT.lua) where the version of a module might not be a string.
* Formalizes the lupa dependency needed for reading lua files.
* packaging changes

* fixed publishing of IPF information, formatted as a PublisherInfo glue2 entity (not defined by glue2 spec), and added it to the modules, compute, activities and services workflows
* fixed parsing of GPU resources on Bridges from scontrol output
* fixed ipf_configure_xsede Location bug: it now correctly puts the configured location into the workflow json file as params