Investigate need for Apple Silicon fix to `Internal_Convertfwrite(...)`

[sdmaclea]

During AppleSilicon initial bring-up we observed a test failure where the Internal_Convertfwrite(...) function was being called with an empty string. This was being treated as a failure during the MultiByte to utf-8 conversion. For the conversion reporting a failure in this case was incorrect. Therefore a fix was included in #40435.

However, it is not clear why this only occurred for osx-arm64 and not osx-x64.

Temporarily remove the fix and investigate the root cause of why this only occurs on arm64. We would like to be sure the fix is not masking another underlying issue.

See discussion https://github.com/dotnet/runtime/pull/40435/files/e186f0106acfd8ebba311581ddbc4e0877442f46#r498157083

Tagging subscribers to this area: @dotnet/ncl
See info in area-owners.md if you want to be subscribed.

[sdmaclea]

One of the tests which showed this issue was readytorun/tests/mainv1/mainv1.sh

{0x425774-0x10056cfcc} ASSERT [CRT    ] at /Users/steve/git/runtime/src/coreclr/src/pal/src/cruntime/printfcpp.cpp.70: 
WideCharToMultiByte failed.  Error is 42424
./mainv1.sh: line 140: 36999 Trace/BPT trap: 5       $CORE_ROOT/crossgen -readytorun -platform_assemblies_paths $CORE_ROOT:`pwd` -out test.ni.dll test.dll

/cc @janvorli

[sdmaclea]

There have been a lot of runtime instability issues where references were getting zeroed. Most seem to be fixed by macOS 11.3 Beta 6. I do not think this early failure in 11.0 Beta is worth investigating.

Closing.