Build failures with newer CUDA Toolchain

[pauleonix]

To get this project to successfully build using CUDA 12.6 one has to uncomment

cuGridder/src/utils.cu

Line 19 in 729db09

//#include <thrust/extrema.h>

and get rid of the unneeded

cuGridder/test/adjointness_1d_test.cu

Line 9 in 729db09

using namespace thrust;

in the test files.

Then the compiler still warns about

cuGridder/contrib/common.cu

Line 184 in 729db09

if (nf < MAX_NF) // otherwise will fail anyway

because of

cuGridder/contrib/common.h

Line 14 in 729db09

#define MAX_NF (int)1e11 // too big to ever succeed (next235 takes 1s)

"floating-point value does not fit in required integral type". So it should probably be replaced with INT_MAX?

[HLSUD]

Really appreciate your comments. Just made a new commit to support cuda 12 except the warnings.