Support Mach-O as a container for R2R images

[agocke]

This has two separate paths: MacOS and iOS. This is because MacOS allows for arbitrary page protection changes (JIT entitlement), while iOS doesn't. That means that MacOS and iOS can use different implementations. Eventually, we should converge to one, the iOS representation, but we can get started by simply using Mach-O as a container of PE images.

• Refactor and include Mach-O object writer from NAOT in R2R (Use an ObjectWriter-based PE writer for crossgen2 #120454)
• Make minimal modifications necessary to load composite R2R images from Mach-O images
  • Add initial doc for future R2R platform-native image (Mach-O, composite only) #120584
  • Extract out minimal concerns from PEImageLayout for R2R images #120777
  • Update our local copy of the Crossgen2 tasks from the SDK to the most recent copy on dotnet/sdk's main #121187
  • Disable PerfMap generation for Apple mobile platforms #121237
  • Add support for emitting Mach-O R2R images #121186
  • Add support for host callback to get information for platform-native composite R2R images #121363
  • Add R2RDump support for Mach-O composite R2R images #121566
  • Enable PlatformNativeR2R test on macOS #121752
• Make modifications necessary for R2R images on iOS
  • Design full Mach-O format
  • Disable optimistic sets when running crossgen2
  • Implement in crossgen2
  • Implement loading in CoreCLR
  • Add runtime test using R2R on iOS
  • Implement loading in the host for iOS

[dotnet-policy-service]

Tagging subscribers to 'os-ios': @vitek-karas, @kotlarmilos, @steveisok, @akoeplinger
See info in area-owners.md if you want to be subscribed.

[janvorli]

While designing this, it would be good to keep in mind that long term, it would be great to also use a native R2R format for Linux.