Allow working with structured CSVs with no header row

Add README with examples

Unit tests for reading and writing unstructured data

Default to reading from / writing to strings for convenience

Rename package prefix

Fix bug writing record fields which collide with impl names (e.g. .size)

Switch to benchmarking with byte arrays instead of disk

Update benchmarking code to warm up and GC before each benchmark

Add options to reader / writer construction

Add benchmarking versus clojure.data/csv

Initial commit: core functionality for fast CSV library