Ensure when USE_BLAS=mkl at mxnet build doesn't break with MKL_USE_ILP64=ON

[access2rohit]

Description

MKL supports int64 indexing if MXNet is build with flag MKL_USE_ILP64=ON but internally mkl uses long long for 64-bit int instead of int64_t. This is fine as long as parameters are passed by value but when they are passed as reference *int64_t is treated as *long hence for 64-bit integer pointers there is a type mismatch by compiler while building MXNet which is cannot convert 'long int *' to 'long long int*' as shown in colsole snippet shown below

../src/api/operator/numpy/linalg/../../../../operator/numpy/linalg/../../tensor/.././c_lapa
ck_api.h: In function 'int MXNET_LAPACK_dgetri(int, int, double*, int, int*, double*, int)'
:                                                                                          
../src/api/operator/numpy/linalg/../../../../operator/numpy/linalg/../../tensor/.././c_lapa
ck_api.h:382:68: error: cannot convert 'long int*' to 'long long int*' for argument '5' to
 'long long int LAPACKE_dgetri(int, long long int, double*, long long int, const long long 
int*)'                                                                                     
       return LAPACKE_##prefix##getri(matrix_layout, n, a, lda, ipiv); \                   
                                                                    ^                      
../src/api/operator/numpy/linalg/../../../../operator/numpy/linalg/../../tensor/.././c_lapa
ck_api.h:388:3: note: in expansion of macro 'MXNET_LAPACK_CWRAP_GETRI'                     
   MXNET_LAPACK_CWRAP_GETRI(d, double)                                                     
   ^~~~~~~~~~~~~~~~~~~~~~~~  

In order to fix it LAPACK and BLAS declarations and calls to MKL need to casted(reinterpret_cast<long long int*>) inside MXNet code base.

[Zha0q1]

https://github.com/apache/incubator-mxnet/blob/master/src/operator/numpy/linalg/np_solve-inl.h#L355
https://github.com/apache/incubator-mxnet/blob/master/src/operator/linalg_impl.h#L1667

Some algorithmic changes are also needed, such as pivot calculation ^

[pengzhao-intel]

@mxnet-label-bot remove [MKL]

[anko-intel]

@mxnet-label-bot add [MKL]