…'t re-export older ac's, and don't overwrite __init__.py (write to an alternate file instead).

…'t re-export older ac's, and don't overwrite __init__.py (write to an alternate file instead).

Write out sources.json in ca, tam, and ad subdirs.

Add skipped test checker to tools

Changed expected result tester to collect all errors instead of throwing an assertion error on the first one.  This makes the test output even more verbose, so I also added a tool that summarizes it, showing for each solution how many scenarios and which tests are failing.

Expected results gives all errors, and tool summarizes them

Buildingintegration progress

Extract 'Low Correction?' entry from excel and return in read_xls(), although it's not written anywhere yet.

Pass bound_correction up from vma extraction (

…uding bound_correction and new description field.

Search for 'Low Correction?' in column Y as well as X.

The second copy of the double-counted adoption must come from ref_base_adoption, not ht, in order to match excel behavior.

Bug fix: ensure that data files have unique names.
Enhancement: delete original data files.

class Scenario has shared class variables that become instance variables when overridden --- but only if you override them.  If you modify them instead, you modify the class variable which then affects all other instances as well.  In this case, it meant that overridden adoption parameters became permanent, and affected subsequent instantiated scenarios.  Detected because walkablecities passed tests by itself, but failed when tested together with other solutions.  I'm amazed that there weren't more issues.

Also deleted a bunch of overrides that were just re-affirming the default value.

bug fix Scenario class

…into globals