Findr

New functions:
Included P-value computation for 4 tests in pijs_gassist_pv and correlation test in pij_rank_pv.
Included using continuous anchors (instead of discrete genotypes) for causal inference in pij(s)_cassist*.
Revisions:
Now allows to load libraries with a different patch version (but the major & minor versions must still remain the same with those of the interface).
Now also tries automatically to load Findr's shared library from ~/.local/lib.

Extreme situation behavior:
Added error checking for few samples (<4).
Added special consideration for extremely skewed LLR during local FDR estimation.
Updated copyright notice.

Interface changes:
Removed trailing _a's for all interface function names.
Removed _tot functions. (See Obsoletes.)
Binary interface has an extra parameter specifying memory usage limit.
Python interface has an optional parameter specifying memory usage limit.

Improvements:
We have introduced the function pij_gassist_trad for traditional inference test.
We have parallelized the remaining part of major computation and obtained even faster speed.
We have modified functions to perform inplace operations as much as possible to save memory.
We have introduced memory usage limit as a parameter (except R interface). Large datasets will be automatically split before calculation to keep memory usage under limit.
We have included p1 calculation for nodiag=1 case.
Better input validity checks for interfaces.
Bug corrections.

Obsoletes:
In order to reduce effort in improving and maintaining the package, we decided to obsolete several non-essential functions. They include:
pij(s)_*_tot functions: they will be hardly needed as more genes can be measured by advancing technologies.
Windows native support: The latest version of Windows has provided support for Ubuntu Bash. All interfaces should integrate better with Bash on Windows. (https://msdn.microsoft.com/commandline/wsl)

This release includes previous patches and supports Windows with MinGW64 from MSYS2. On Mac and Linux, installing from the source would be more convenient, while on Windows, we recommend using the pre-built binary release. The binary release can be installed by right clicking on install.cmd and choosing run as administrator.