In some data processing tasks we need to use huge maps or sets that are bigger
than available JVM heap space or they are loading too slow to standard Java or Scala Maps.
We use TSV format (text file with tab separated columns) for persist this kind of Maps or Sets.
Some columns are used as a key and rest of columns as a value.
Idea of this library is simple. We can prepare these maps once (sort by key),
store it to file and then use it as memory mapped file.
Searching key in sorted file has log(n) complexity.
If more processes uses the same memory mapped file, it exists in memory just once (on Linux).
This file can be loaded lazy by OS.