Remove built-in COM interop support from Mono runtime

[jkoritzinsky]

This support is always disabled, so let's delete it.

Fixes #97703

[lambdageek]

Thanks!

[jkoritzinsky]

Looks like I need to make the AOT compiler resilient to encountering a [ComImport] type. I'll take a look tomorrow at that.

[lambdageek]

Linux arm64 libraries test crash is possibly related:

Thread 11 (Thread 0xffffb2e620e0 (LWP 34) ".NET Long Runni"):
#0  0x0000ffffbd926800 in __GI___wait4 (pid=pid@entry=35, stat_loc=stat_loc@entry=0xffffb2e5d450, options=0, options@entry=-1293561008, usage=usage@entry=0x0) at ../sysdeps/unix/sysv/linux/wait4.c:30
#1  0x0000ffffbd926758 in __GI___waitpid (pid=pid@entry=35, stat_loc=stat_loc@entry=0xffffb2e5d450, options=options@entry=-1293561008) at ./posix/waitpid.c:38
#2  0x0000ffffbd4c147c in dump_native_stacktrace (signal=<optimized out>, mctx=<optimized out>) at /__w/1/s/src/mono/mono/mini/mini-posix.c:843
#3  mono_dump_native_crash_info (signal=<optimized out>, mctx=mctx@entry=0xffffb2e5deb0, info=<optimized out>) at /__w/1/s/src/mono/mono/mini/mini-posix.c:870
#4  0x0000ffffbd472218 in mono_handle_native_crash (signal=0xffffbd31027b "SIGSEGV", mctx=0xffffb2e5deb0, info=0xffffb2e5e210) at /__w/1/s/src/mono/mono/mini/mini-exceptions.c:2967
#5  0x0000ffffbd3dfb1c in mono_sigsegv_signal_handler_debug (_dummy=11, _info=0xffffb2e5e210, context=0xffffb2e5e290, debug_fault_addr=0x20) at /__w/1/s/src/mono/mono/mini/mini-runtime.c:3888
#6  <signal handler called>
#7  mono_object_handle_isinst (obj=..., klass=klass@entry=0x0, error=error@entry=0xffffb2e5f4e0) at /__w/1/s/src/mono/mono/metadata/object.c:6807
#8  0x0000ffffbd5f212c in mono_object_isinst_checked (obj_raw=0xffffbca2a1e0, klass=0x0, error=error@entry=0xffffb2e5f4e0) at /__w/1/s/src/mono/mono/metadata/object.c:6790
#9  0x0000ffffbd5c2ee4 in mono_marshal_isinst_with_cache (obj=0xffffbca2a1e0, klass=0x0, cache=0xffff9005a820) at /__w/1/s/src/mono/mono/metadata/marshal.c:4445
#10 0x0000ffffbcd6ff7c in ?? ()
#11 0x0000ffff94003580 in ?? ()

Note mono_marshal_isinst_with_cache is getting called from JITed code with a NULL klass` argument.

[jkoritzinsky]

@lambdageek There's a known issue for that failure #90019.

Looking at one of the logs from one of the other instances, it looks like the same symptom

#6  <signal handler called>
#7  mono_object_handle_isinst (obj=..., klass=0x0, error=0xf15fb198) at /__w/1/s/src/mono/mono/metadata/object.c:6884
#8  0xf77991e8 in mono_object_isinst_checked (obj_raw=0xf6a6ece0, klass=0x0, error=0xf15fb198) at /__w/1/s/src/mono/mono/metadata/object.c:6867
#9  0xf7763eb8 in mono_marshal_isinst_with_cache (obj=0xf6a6ece0, klass=0x0, cache=0xf1728fb0) at /__w/1/s/src/mono/mono/metadata/marshal.c:4356
#10 0xf6df0220 in ?? ()

As you're more familiar with this code than I, I'll defer to you.

[lambdageek]

Oh yes, that's the same crash. I didn't recognize the test suite. I think of 90019 as somehow being related to generic collections (perhaps some kind of generic sharing race condition). But yes, it's the same issue.