1, edited README.md, add link to build example file.
fix bugs.
1) Fix Eigen Sparse Matrix calculating error when the matrix is 0 by 0 of function erase_B_and_Z
The error of function erase_B_and_Z only show up when the size of selected snp is 1
and pJ of this snp is greater than _jma_p_cutoff. And then go to the step
which remove this snp's data from _B, _B_N, _B_i, _Z etc(function
erase_B_and_Z), and then the error was triggered.
For example(direct cause of error):
```
SimplicialLDLT<eigenSparseMat> ldlt_B(_B);
_B_i = ldlt_B.solve(_B_i).eval();
```
if the _B is 0 by 0, then above sentence will induce an error.
The error look like this in gdb
```
__memmove_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:416
416 movq -8(%rsi, %rdx), %rcx
```
2) Fix error of function slct_entry when selected snp is 0;
If the size of selcted snp is 0 and remained is greater than 0, the
slct_entry step will going on. But the size of _B_N, _Z_N is 0, and
which will cause same calculating error of Eigen matrix or vector as
1). So we need add a new snp to initiate those matrix. We using snp
among remained snp which have biggest chisq value, and using this
snp to re-initiate the variable(_B_N, _Z_N etc).
1, edited README.md, add link to build example file.
fix bugs. 1) Fix Eigen Sparse Matrix calculating error when the matrix is 0 by 0 of function erase_B_and_Z
2) Fix error of function slct_entry when selected snp is 0;
Fang Hailing, Ma Ruilei, Qi Ting