csvjoin: Implement case-insensitive key comparison

[anorth]

This is much nicer than lowercasing files to do the join then trying to restore case that's important later on.

Thanks for the great tools!

[anorth]

This is for a virtualenv, but let me know if you have another convention you strongly prefer.

[coveralls]

Coverage decreased (-0.1%) to 87.099% when pulling a4f93c7 on dssrt:master into ffe5f44 on wireservice:master.

[jpmckinney]

Instead of making this an option that only applies to csvjoin, I think it we could use a global option to normalize headers. This relates to #525, which needs to normalize headers to be SQL-safe.

[fgregg]

@jpmckinney this PR doesn't actually seem to be about headers?

[jpmckinney]

@fgregg You're right - not sure what I was thinking in 2016 😅

[jpmckinney]

@fgregg Is this the sort of think that csvlink does? I'd be happy to add a note to csvjoin, instead of adding lowercase (which i just one possible transformation to assist in linking tables).

[fgregg]

csvlink is not a good fit for this. one thing i’ve often wished for is passing a little code into csvpy on the command line that could make this really composable

something like csvpy -c identifier -p “d.lower()” example.csv