More graceful support for long column names

_From @mbauman on December 21, 2016 22:42_

Right now it's easy to construct column names that become too long and get truncated. Truncated column names can easily become exact duplicates. Exact duplicate column names throw errors.

_Copied from original issue: dssg/collate#31_

The issue names no files or tests; start by locating the code that constructs and truncates column names, then reproduce a collision from two names that share a truncated prefix. Done means long names no longer silently become duplicate columns or trigger the reported duplicate-name error, with a regression test if the project provides one.

Written by the indexing model from the issue text.