Transforming multiple files concurrently
At present, the application supports file input, but only from a single file. A valid use case could be to provide multiple files and create a new file with the transformed data for each one. Given the transformation logic is IO-bound, doing this concurrently makes sense and should lead to better performance.
The reason why IO-bound logic and concurrency go so well together is because of the Crystal scheduler. When a fiber gets to a point in its execution where it is dependent on some piece of data from an IO, the scheduler is able to seamlessly put that fiber to the side until that data has arrived.
A more concrete example of this in action would be to look at how the standard library's HTTP::Server functions. Each request is handled in its own fiber. Because of this, if another HTTP request needs to be made during the processing of a request, such as to get data from an external API, Crystal would be able to continue to process...