Update release instructions with regenerate SPGO data step

[Alan-Jowett]

SPGO data makes a massive difference in the BPF performance benchmarks.

Before SPGO:

C:\Users\Administrator\Desktop\bpf_performance>Release\bpf_performance_runner.exe -i tests.yml -e .sys -t "bpf_tail_call" -c 100000000
Timestamp,Test,Average Duration (ns),CPU 0 Duration (ns),CPU 1 Duration (ns),CPU 2 Duration (ns),CPU 3 Duration (ns),CPU 4 Duration (ns),CPU 5 Duration (ns),CPU 6 Duration (ns),CPU 7 Duration (ns)
2024-11-06T17:45:06-0800,bpf_tail_call,117,114,120,118,117,117,121,119,111

After SPGO:

C:\Users\Administrator\Desktop\bpf_performance>Release\bpf_performance_runner.exe -i tests.yml -e .sys -t "bpf_tail_call" -c 100000000
Timestamp,Test,Average Duration (ns),CPU 0 Duration (ns),CPU 1 Duration (ns),CPU 2 Duration (ns),CPU 3 Duration (ns),CPU 4 Duration (ns),CPU 5 Duration (ns),CPU 6 Duration (ns),CPU 7 Duration (ns)
2024-11-06T17:30:28-0800,bpf_tail_call,60,62,60,58,57,60,60,63,65

But SPGO counts can quickly become stale, resulting in fairly large performance regressions.