Enable updated Accelerate headers for Darin kernel >= 22.4

[PaulWessel]

Getting tired of the tantrum Xcode has added if we call the deprecated dgemm_ without including the new headers via compiler flag -DACCELERATE_NEW_LAPACK. This PR has ben following changes:

1. cmake/ConfigUserAdvancedTemplate.cmake: Adds -DACCELERATE_NEW_LAPACK to the compiler flags if we are on Darwin >= 22.4
2. CMakeLists.txt: Determine if we are on version 22.4 and if so set Cmake variable NEW_ACCELERATE_LAPACK to 1 (else 0).
   3. gmt_vector.c: Use #ifdef ACCELERATE_NEW_LAPACK to cast our parameters using Apple's new dgemm_ definition.

This PR now compiles quietly for me on macOS Sonoma. Ideally:

• Someone who knows Cmake better than me (@seisman, @remkos ?) should see if the new code in the advanced template can be simplified.
• Give it a try if you have an older kernel since I am unable to test that.
• Also build on non-macOS to ensure nothing funny happened.

[remkos]

As you have notices, I needed to make some changes:

• Renaming NEW_ACCELERATE_LAPACK to ACCELERATE_NEW_LAPACK did not have the result I expected, though I should have realised that. Nonetheless, consistency is practical.
• Putting stuff in ConfigUserAdvancedTemplate.cmake did not make much sense. I would not have looked into the template myself. It should be part of the CMakeList hierarchy.
• The whole stuff needed to be moved to src/CMakefile.txt to have effect, because only there it is determined if LAPACK is even there.

[remkos]

It works now for me.
@PaulWessel and @seisman please check if this works for you.
It should probably work "out of the box" with a simple make. Otherwise redo a cmake in your build directory, then make.

[PaulWessel]

I guess I copied the new stuff from that template to my own advanced cmakefile and as far as I can tell it found everything and compiled as expected. So I think it works IF you update that file.

I agree that users on recent macs who already have GMT via git will not notice the change since they may have an advanced file that is not getting updated. I can look at this tomorrow pm unless you do something first.

[remkos]

It works now for me without having touched my ConfigUser.cmake, which is what I wanted to achieve, and as you pointed out is what users would likely expect. We do not need to bother them with switching logic in their user configuration.

I have nothig to change. Just wait for @seisman and you to see if it works for you.

[PaulWessel]

Great, works fine for me - yours is the right solution. All tests still pass that passed before.