r.accumulate: Add parallelization

[shreyasudaya]

OpenMP parallelization added.
To do:
Test timing and parallelization. This along with r.horizon is where I am confused about testing.

[HuidaeCho]

@shreyasudaya Have you actually compiled and tested your parallelized code? Parallelization was not in my mind at all when I first wrote this module and I'm not so sure if simply adding parallel for loops will correctly parallelize all the algorithms. Just tested this PR and I wasn't able to compile it:

calculate_lfp_iterative.c: In function ‘calculate_lfp_iterative’:
calculate_lfp_iterative.c:79:52: error: ‘j’ undeclared (first use in this function)
   79 | #pragma omp parallel for schedule(dynamic) private(j)
      |                                                    ^
calculate_lfp_iterative.c:79:52: note: each undeclared identifier is reported only once for each function it appears in

j is declared inside the for loop and isn't visible to OMP outside of it.

[HuidaeCho]

To be honest, I don't think it's straightforward to parallelize this module because many things are not thread-safe. One easy example is push_up(). When multiple threads try to allocate memory at the same time within trace_up(), a race condition will occur. Parallelization of any algorithms in this module would require a lot more work, if possible at all, and, unfortunately, I have to suggest closing this PR.

[petrasovaa]

@HuidaeCho, would you have any suggestion for a different tool that could be easier to implement parallelization for? I don't think there are many low hanging fruits, but maybe v.surf.idw?

[HuidaeCho]

Closing this PR.