Add support for properly disabling the DLONG support

[imciner2]

This fixes the compile error on systems where QDLDL_int is a 32-bit integer instead of a 64-bit integer. The problem is that the DLONG flag was being always defined in the code, and that then the qdldl interface code was always calling the long version of the amd_order instead of the one for the size of the integer. I have tested this on a 32-bit Beaglebone Black and my 64-bit Fedora machine and it compiles on both. (I can't run the tests on the 32-bit machine because I don't have Python >3.7).

Note that this will depend on PR osqp/qdldl#37 being merged into QDLDL and then being included here. In the mean time, it can be tested with the proposed QDLDL fix by running the following commands inside the qdldl submodule

git remote add imremote https://github.com/imciner2/qdldl.git
git fetch imremote
git checkout im/confheader

Fixes #15

[tatokis]

If anyone is interested, the patch should now be

diff --git a/c/amd/include/SuiteSparse_config.h b/c/amd/include/SuiteSparse_config.h
index ea3cea4..5a15deb 100644
--- a/c/amd/include/SuiteSparse_config.h
+++ b/c/amd/include/SuiteSparse_config.h
@@ -43,7 +43,6 @@ extern "C" {
 #endif
 
 #include "qdldl_types.h"
-#define DLONG
 typedef QDLDL_float c_float;
 
 #include <limits.h>
diff --git a/c/amd/include/amd_internal.h b/c/amd/include/amd_internal.h
index eccde9f..cd745fd 100644
--- a/c/amd/include/amd_internal.h
+++ b/c/amd/include/amd_internal.h
@@ -38,9 +38,6 @@
 #endif
 
 #include "qdldl_types.h"
-#ifndef DLONG
-#define DLONG
-#endif
 
 /*
     To enable debugging, uncomment the following line:
diff --git a/cpp/qdldl.cpp b/cpp/qdldl.cpp
index e304dbe..4751926 100644
--- a/cpp/qdldl.cpp
+++ b/cpp/qdldl.cpp
@@ -1,5 +1,6 @@
 #include "qdldl.hpp"
 
+#include "qdldl/include/qdldl_types.h"
 using namespace qdldl;
 
 
@@ -31,7 +32,11 @@ Solver::Solver(QDLDL_int n, QDLDL_int * Ap, QDLDL_int *Ai, QDLDL_float * Ax){
 	Pinv = new QDLDL_int[n];
 
 	// Permutation
+#ifdef DLONG
 	QDLDL_int amd_status = amd_l_order(nx, Ap, Ai, P, NULL, NULL);
+#else
+	QDLDL_int amd_status = amd_order(nx, Ap, Ai, P, NULL, NULL);
+#endif
 	if (amd_status < 0)
 		throw std::runtime_error(std::string("Error in AMD computation ") + std::to_string(amd_status));