Use Profile-Guided Optimization (PGO) to speedup programs (e.g. migrations) by up to ~14%

[fxamacker]

The Go compiler supports profile-guided optimization (PGO).

• Go 1.20 - PGO is a new preview feature
• Go 1.21 - PGO is ready for general use and improved
• Go 1.22 - PGO is ready for general use and further improved

Go release notes mentioned 2-14% speedup by using PGO but it might be better than that in some cases as it improves.

Updates these (and maybe others):

• Epic - Migration optimization #3297
• [Execution State] Memory usage and performance optimisation flow-go#1744

I opened this here in onflow/cadence because Epic #3297 mentions hardware, multiple machines, etc. to speedup migrations.

[turbolent]

Now that we're on Go 1.22 we should be able to try this out more easily

[fxamacker]

Now that we're on Go 1.22 we should be able to try this out more easily

@turbolent BTW, thank you for pushing for and bumping Go to 1.22. 👍

[turbolent]

Applying PGO to the state extraction resulted in a ~2% improvement. See https://www.notion.so/flowfoundation/State-Migration-Result-Tests-01e501fceb8444cca8a598d6111f2e3c?pvs=4#f1464668c5324c0a8568a47e8b02edf6 for details.

Though the result is not that impressive, given that this is basically a free win, we should still consider building the binaries with PGO enabled when we run the migrations. cc @zhangchiqing @j1010001