Explicit encoding handling?

[nrinaudo]

Encoding currently relies on implicit values of scala.io.Codec being in scope, which can be unpleasant.

First, there's always an implicit value in scope, the system's default, which can cause unexpected bugs - typically when testing on one OS and running on another.

Second, it can make some use cases needlessly convoluted. Trying to write some code that reads a CSV file in one encoding, does some sort of processing, and writes it back to CSV in another encoding is much more frustrating than it should be.