With the ansible configuration centralized into a top level torque-sites
directory (roles), the majority of the tasks main.yml and supporting
files can be removed with the roles being added to the top level
playbook.

This required a change to the base roles:
* Board Members from Okta are Decision Makers

With the ansible configuration centralized into a top level torque-sites
directory (roles), the majority of the tasks main.yml and supporting
files can be removed with the roles being added to the top level
playbook.

This required a change to the base roles:
* torquedataconnect_sheet_name can be different from the competition

With the ansible configuration centralized into a top level torque-sites
directory (roles), the majority of the tasks main.yml and supporting
files can be removed with the roles being added to the top level
playbook.

This required a few changes to the base pipeline:
* Add composer to mediawiki install
* Composer version update
* Add postgres base package
* Add LFC Evaluators as a known SAML group
* Add pre-req installation for mediawiki

Then removing all the old stuff from 100Change2017 and updating
the yml file to reference the general stuff.

Clean it up so things are a bit more declarative, as well as moving the
simpelsaml group config code over to the permissions, so that the
separation of concerns is a bit more straightforward.

Issue #62: Create security model for torque-sites
Issue #64: Pluralize the names of the abstract user groups

A lot of the configuration was inline blocks that were dumped directly
into LocalSettings.php.  This change separates those configurations out
into their own files for easier maintenance and documentation.

Issue #59: Change MediaWiki configurations to have multiple files

For centralizing roles, Climate2030 being the most recent competition
was the best candidate to base the refactoring on.  The refactoring took
the form of splitting out the main.yml from the mediawiki task into many
main.yml files in different roles.

Because roles are getting centralized, the documentation needs to be
updated with how and how to configure it.

Some of these changes actually got applied to all competitions because
they represent categories of changes that we want to prevent in general:

* All mediawiki titles should be capitalized
* '' in data should be replaced by "
* Evaluation comments should have fix_cell applied to them
* A new processor for <BR>s should be added, to remove them
  - This is due to some cells having new lines where they shouldn't

In addition, the following changes were needed in LoneStar2020

* A corrections file for a title that was incorrect
  - Needed a corrections file instead of edit interface because title's
    are used at a system level

Also requires an update to the ots subversion directory

Along with a change in the svn bigdata directory for EO2020, adds the
attachments for the final 5 proposals.

Change the main proposals list to a table using the table formatter.

This stemmed from a need to have two different kinds of TOCs, one that's
a simple list, and one that's a table.  This required rearchitecting a
little bit and abstracting out how the tocs are built, with two initial
formatters present.

Included adding the LFC Analysis and Evaluation pages to the wiki (which
required an ansible change)

This is based on a partial sheet, so we couldn't use the normal
evaluation etl processing, but that will be coming in the next few
weeks.  Also got a set of ranking data for which proposals should get
lfc analysis pages.

Add no-pare option

Mostly documentation improvements.

Required some updates to the etl pipeline:
* Creating the LFC Columns
* Adding a ranking for within a solution category
* Creating the evaluations and analysis pages

And some ansible changes
* Adding print css to make sure lfc pages print out correctly
* Adding the hook to create tabs for evaluations
* Updating permissions to give sysops print ability

This was moved to the central etl pipeline as part of the etl refactor,
and this file in LoneStar2020 was missed.

Documentation about option included in the documentation in the diff..
There's also a small typo fix where -a was in the usage instead when
the option is -c

It was proving too annoying with updates as we change the data, and it's
not super important to maintain this sha anyway.  We can reinsitute when
we get demo data into a fully passed review place.

This prevented rendering correctly on target wiki.

53 refactor etl

Now uses the proper python sorting with comparison functions on the
Attachment object, with a default rank of None that gets ordered to the
bottom.  The old rank of 99 was both broken (the sort was written
improperly), and weirdly a magical number.

From review #77: Refactor ETL code to create central utilities

This passes through to the compose-and-upload files

For the former, we want to allow uploads to be faster for development.
For the latter, pare is now a standard argument, while ignoring the
second line is a very uncommon, so we want them to be in order of use.

From review #77: Refactor ETL code to create central utilities

As organizations evolve, so must the code.

From review #77: Refactor ETL code to create central utilities

From review #77: Refactor ETL code to create central utilities

Because the script wasn't unpacking the attachments when
SKIP_ATTACHMENTS was on, it wouldn't properly create the spreadsheet
that had the attachment information.  That created a faulty upload.

From review #77: Refactor ETL code to create central utilities

From review #77: Refactor ETL code to create central utilities

From review #77: Refactor ETL code to create central utilities

Each of the compose files came from an original script that was very
well documented, and that documentation transformed and moved into
the top level README's and whatnot, where it should be.  It also became
out of date and confusing.

From review on #53: Refactor ETL code to create central utilities