More ValueOption in compiler: part 3 (and the last)

[psfinaki]

Closes #16286

If my RegEx is correct, this should be it.

---

Some benchmarks (for what was easy to bench):

Before:

Method	Mean	Error	StdDev	Gen0	Allocated
TaskCancelled_TargetPattern	9.627 ns	0.4100 ns	1.2024 ns	0.0004	24 B
TaskCancelled_OtherPattern	4.766 ns	0.4129 ns	1.2173 ns	-	-
StopProcessing_TargetPattern	11.123 ns	0.6545 ns	1.9297 ns	0.0004	24 B
StopProcessing_OtherPattern	4.433 ns	0.2699 ns	0.7957 ns	-	-
ConstToILFieldInit_TargetPattern	15.940 ns	0.9367 ns	2.7472 ns	0.0008	48 B
ConstToILFieldInit_OtherPattern	2.418 ns	0.0852 ns	0.1621 ns	-	-

After:

Method	Mean	Error	StdDev	Gen0	Allocated
TaskCancelled_TargetPattern	0.0010 ns	0.0037 ns	0.0066 ns	-	-
TaskCancelled_OtherPattern	3.5823 ns	0.2947 ns	0.8688 ns	-	-
StopProcessing_TargetPattern	2.2672 ns	0.1675 ns	0.4938 ns	-	-
StopProcessing_OtherPattern	2.9171 ns	0.2328 ns	0.6863 ns	-	-
ConstToILFieldInit_TargetPattern	9.8902 ns	0.4454 ns	1.2707 ns	0.0004	24 B
ConstToILFieldInit_OtherPattern	3.0223 ns	0.1553 ns	0.4432 ns	-	-

[github-actions]

❗ Release notes required

---

✅ Found changes and release notes in following paths:

Change path	Release notes path	Description
src/Compiler	docs/release-notes/.FSharp.Compiler.Service/8.0.400.md

[psfinaki]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[vzarytovskii]

We'll have to wait with this one for now, since now we're having some live issues with SO

[vzarytovskii]

I've merged couple of fixes and workarounds for the SOs in CEs and in big nested expressions. We can probably merge this one (extra would be to test AnyCpu variant on those).

[abelbraaksma]

These performance improvements are quite impressive and more than I'd expect from such change. Nice!

[psfinaki]

I ran the current version of fscAnyCpu.exe against the code from this and this PRs, don't notice any regressions there.