Add debugging tips

[stgeke]

• How to attach the debugger to running nekrs job
• How to print stacktraces when the job hangs
• How to correctly identify if a kernel failed (CUDA_LAUNCH_BLOCKING=1)?
• Useful tools: valgrind, compute-sanitizer
• Potential workarounds: UCX_MEMTYPE_CACHE=n, OMPI_MCA_io=romio341 etc.

[yslan]

We have some starting templates here:
https://nekrs.readthedocs.io/en/latest/problem_setup/running.html#runtime-control-with-signals
https://nekrs.readthedocs.io/en/latest/problem_setup/debugging.html

It's also worth noting tracing requires -DENABLE_CPPTRACE=on