Don't duplicate existing rooms when importing schedule CSV

* Old implementation needs to see exactly the same rooms in exactly
the same order every time it loads new data, otherwise it will
create a duplicate entry for the room that differs only in display
order.
* New implementation ignores the display order when checking to see
if the room already exists.

Wrap input CSV in an TextIOWrapper

Closes #24

Taking one out of PyCon's (US) book

We're lock-step with symposion, and upstream is dead.
Vendor it.