Initial hot cold splitting support for crossgen2/VM

[cshung]

This PR marks the start of the work to get crossgen2 to support hot/cold splitting.

Here are the changes:

1. Added a section named Scratch, to be described in (13) below for more detail.
2. Create the _methodColdCode from _coldCode and _coldCodeRelocs during PublishCode.
3. Null _methodColdCode and empty _coldCodeRelocs during CompileMethodCleanup.
4. Create _methodColdCode during allocMem.
5. Change reserveUnwindInfo to ignore unwind reservation for coldCode.
6. Change allocUnwindInfo to ignore unwindInfo allocation for 0 bytes.
7. Change findKnownBlocks to return _coldCodeReloc when BlockType.ColdCode is asked for.
8. Change recordRelocation to set relocTarget to _methodColdCodeNode when BlockType.ColdCode is asked for.
9. Implement MethodColdCodeNode

• Storing the cold code.
• Make sure sorting will place cold code blocks after hot code blocks.
• Make sure cold code blocks are ordered the same as the hot code blocks.

10. Change MethodGCInfoNode to

• Emit the chained unwind info for the cold code block, after the hot code GCInfo, so the hot and cold code unwind info are currently mixed.
• Changed CalculateFuncletOffsets so that the offset for the cold code chained unwind info is included if it exists.

11. Changed MethodWithGCInfo to:

• Have a reference to the MethodColdCodeNode.
• Changed GetFixupBlob to include relocation for the cold code.

12. Changed RuntimeFunctionTableNode to:

• Emit the RuntimeFunction entries for the cold code block after the hot code blocks.
• Prepare the cold to hot code mapping for ScratchNode.

13. Implement ScratchNode - the session is simply an array of cold code runtime function index to hot code runtime function index, so all even entries (e.g. 0, 2, 4) are cold code runtime function index and the corresponding odd indexes (e.g. 1, 3, 5) are the corresponding runtime function index for the corresponding hot code. The array is sorted by the cold code indexes, by (9), the hot code indexes should be sorted as well.
14. Implemented the --hot-cold-splitting command-line option, and turn on the CORJIT_FLAG_PROCSPLIT flag as needed.
15. Implemented the various code manager contracts:

• JitCodeToMethodInfo already gave us a runtime function index, we do not know if it is hot or cold, so we use the scratch table to find it. If we got an even index, that would be cold code and therefore we map it back to the hot one.
• Change GetFuncletStartOffsets to return the cold code block as well - technically the cold block is not a funclet, but it appears that the debugger might want it.
• Override IsFunclet so that it avoids returning true for the cold code block, by searching the scratch table.
• Change JitTokenToMethodRegionInfo so that it will find and return the cold code block - again, by search the scratch table since we already have the hot part runtimeFunction.
• Change DebugDebugger so that it uses EECodeInfo to compute relative offset instead of a simple subtraction, that will allow us to account for the cold code.

This work is far from complete - here is a list of issues filed for the project - these are known issues for now:

• [Hot Cold Splitting] Generate the cold code in the proper session for Unix. #1901
• [Hot Cold Splitting] Pack the cold unwind data together #1902
• [Hot Cold Splitting] Redesign the hot cold mapping data structure to allow fast lookup #1903
• [Hot Cold Splitting] Optimize the exception processing to avoid repeated hot-cold mapping lookup  #1904
• [Hot Cold Splitting] GCStress testing #1905
• [Hot Cold Splitting] Conditional hot-cold mapping generation. #1906
• [Hot Cold Splitting] Make sure the debugger work under hot cold splitting #1907
• [Hot Cold Splitting] Fix R2RDump to support dumping hot-cold parts together #1908
• [Hot Cold Splitting] Fix R2RDump to dump the hot cold map #1909
• [Hot Cold Splitting] Fix ILSpy.ReadyToRun disassembling #1910
• [Hot Cold Splitting] Performance measurement #1911
• [Hot Cold Splitting] Avoid relying on class code value  #1912
• [Hot Cold Splitting] Ensure SOS works with hot/cold splitting #1913

@amanasifkhalid, @BruceForstall

@dotnet/crossgen-contrib

[BruceForstall]

This is a good start. Let's get it merged so we can enable collaboration.

[BruceForstall]

Added #1913

[BruceForstall]

Have you followed all the instructions on https://github.com/dotnet/runtimelab/blob/docs/CreateAnExperiment.md to make this a "proper" runtimelab feature branch?

[cshung]

I merged the PR for now, feel free to work on top of it.
In the meantime, I will follow up with @hoyosjs to make sure the branch is set up correctly.
In case it isn't, there might be some churns to the actual commits, but it shouldn't impact the code at all.

[trylek]

I have added some feedback that should be ultimately addressed by the final implementor of hot-cold code splitting in Crossgen2 (Eugenio); in general the code changes seem reasonable to me, thanks Andrew for providing our intern with such a great head start!

[hoyosjs]

I merged the PR for now, feel free to work on top of it. In the meantime, I will follow up with @hoyosjs to make sure the branch is set up correctly. In case it isn't, there might be some churns to the actual commits, but it shouldn't impact the code at all.

Looked at all things I can think of - looks fine to me

[MichalStrehovsky]

I merged the PR for now, feel free to work on top of it. In the meantime, I will follow up with @hoyosjs to make sure the branch is set up correctly. In case it isn't, there might be some churns to the actual commits, but it shouldn't impact the code at all.

Looked at all things I can think of - looks fine to me

The instructions also suggest deleting .github/CODEOWNERS. It's the reason why I was pinged for review (I don't mind that part). But it will also cause a lot of people to be pinged for review if you ever do a merge pull request from runtime main branch (because those touch a lot of files). Deleting it will avoid bothering those people. Doesn't apply if you don't intend to merge from main.