ITS: use faster atan2 for trackleting

[f3sch]

Instead of using the full atan2 which is rather expensive one can use a faster approximate method.
The speedup on cpu is around 10% for the trackleting phases (on cpu this is 5% of the total time with 20 threads, so a gain of around 0.5%)
On gpu the speedup is much less 0.5% (prob. within fluctuations, on gpu this phase is around 26% of the total time with <<<30,256>>>, so only 0.13% faster).

The difference in output is minimal (below per-mille). Below is the comparison. The plot in the bottom right shows that after switching with deterministic mode the gpu part still 1:1 reproduces the cpu output.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5