Xcode 13 support and `standard` row delimiter

• Xcode 13 has been supported by raising the iOS requirements to iOS 11.
  The library still supports iOS 7+, but the SPM now defines iOS 11+. If you want to support older OSes, modify the SPM manifest.
• Support for more than one row delimiter (e.g. for the standard \r\n and \n) at the same time.
  CSVReader will end a row when one of the row delimiters is encountered. Use just one row delimiter for better performance.

Stricter parsing

• CSVReader and CSVDecoder are now "stricter" while parsing CSV data.
  Now errors are thrown when encountering ill-formed rows, instead of just finishing the parsing process (#34).
• CSVReader and CSVDecoder ignore empty lines.
  Usually encountered at the end of a file.
• Minor enhancements on documentation.

Second batch of performance enhancements

• CSVWriter encoder is faster for UTF8 encodings.
• Shift-JIS encoding is now supported (#29).
• CSVEncoder and CSVDecoder have experienced major speed ups, due to drops on unnecessary retains/releases.

Optimizations & decodeIfPresent

• The first batch of optimizations have landed.
• Fix for floating-point encoding/decoding (#20)
• Change decodeNil() behavior to more closely follow the documentation.
• decodeIfPresent can now be safely used.

Lazy Row Encoding

• Both CSVEncoder and CSVDecoder now support lazy functionality.
• The lazy API has been renamed to be as similar as possible in the encoder/decoder.
• README has been expanded with more examples.

Top Level Codecs & InputStream Support

• CSVEncoder/CSVDecoder adopts TopLevelEncoder/TopLevelDecoder when Combine is present.
• CSVReader and CSVDecoder now also accept InputStreams.
  The introduction of this feature allows easier usage of CodableCSV by Command-line applications reading information from the .standardInput (i.e. stdin).
• Most errors thrown by CodableCSV functions are now CSVErrors.
  All CSVErrors include the failure reason and provide help cues to avoid said problem.
• Documentation has been expanded to cover all public and internal functionality.

Lazy Row Decoding

• CSVDecoder now supports on demand (lazy) decoding.
• New encoding/decoding configuration strategies have been added.

CSVEncoder, Buffer Strategies & API Refining

• A full-fledge CSVEncoder has finally been implemented.
  Full support for Codable has been achieved. You can now use keyed, unkeyed and single value containers when neeeded.
• CSVEncoder and CSVDecoder support for different buffering strategies.
  This translates in less memory usage for sequential or assembled runs.
• CSVReader/CSVWriter API have been renamed to match CSVEncoder/CSVDecoder API.

Linux Support & Custom Escaping Scalars

• Custom escaping scalars are supported thanks to @josh (#13).
  This includes the ability to disable escaping functionality on parsing or serializing CSVs.
• Linux is officially supported.
  All tests now also run on Linux (Ubuntu 18.04) through Github actions.
• Trim strategy now throws an error at initialization when it contains delimiter characters or escaping scalars.
• The repo now provides not only a high-level roadmap, but also a detailed plan on which features are being worked next.

Greater Performance & Easier Debugging

• Expand input/output support to Data, String, and files (through URLs).
• Reimplemented CSVReader and CSVWriter for greater performance.
• Introduction of CSVError adopting SE-112 protocols for easier debugging.
• Make Decoder fully immutable.
• Expand tests on CSVReader, CSVWriter, and CSVDecoder.
• OS requirements reduced to macOS 10.10, iOS 8, tvOS 9, watchOS 2.
• First trials on Linux.
• Fixed bug on trim character strategy.