ENH: Allow referencing of Excel columns by their Excel column names.

[snth]

I find it more convenient to refer to the columns as they are in Excel by their string representations.

The excel2num() function I found on StackOverflow. Does that need to be referenced/acknowledged?

I'm happy to put in some unit tests but couldn't find a natural home for it. Most of the ExcelFile tests sit in test_frame.py but I think something like test_io.py or test_parsers.py would be a better fit.

[wesm]

All you have to do is push additional commits to the branch (well, your master branch in this case-- see http://pandas.pydata.org/developers.html#working-with-the-code for the future). Would you mind adding a unit test that addresses this new behavior (you can use one of the .xls files that are already used in the test suite)

[snth]

Thanks Wes. I wanted that feature for myself so I just started hacking at it. I will follow the guidelines in future.

I've made the change I pointed out above and added a unit test for the 'str' behaviour that replicates the existing parse_cols unit tests.

Please review and see if you're happy with it.

[wesm]

cherry-picked the relevant commits. thanks!