Version 0.0.5 released.

• Revert UTF-16 and UTF-32 label change:
  it was an error to specify endianness for texts with BOM.
  The Unicode standard explicitly warns against it, and it actually
  even (partially) breaks conversions.
• Added supports:
  • French: Windows-1252.
  • German: ISO-8859-1, Windows-1252
  • Esperanto: ISO-8859-3
  • Turkish: ISO-8859-3 and ISO-8859-9
  • Thai: ISO-8859-11 (and TIS-620 model rebuilt).
• Single Byte charset detection algorithm improved:
  detection of control characters lowers confidence.