CMake support, performance improvements

[Deamhan]

There is no new features or format changes in my code changes, my main goals is performance and portability. Thus I:

• created CMakeLists.txt that works well with Visual Studio and QtCreator (tested)
• profile guided optimization works well on Visual Studio
• reduced stack consumption (now all stack frames less than 4K)
• improved "hot" functions inlining
• improved exception safety
• improved I/O performance
• modernized code according current standarts
• eliminate exceptions to improve performance and compatibility with C code

As result time of running VS2017 generated binary (x64, Release) in my test (2 files over 300 MB, --all-in-ram mode) reduced by 30% (compared with original code), with PGO result is 35%. Performance improvement on GCC (8.3.1) about 10%, but it is better than nothing.