JIT: stop creating gc poll blocks in morph

[AndyAyersMS]

Instead, mark the block as needing a GC poll and add one later in the normal GC poll insertion phase. This only impacts x86.

This change also stops creating inline GC polls for these cases in unoptimized code.

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Instead, mark the block as needing a GC poll and add one later in the normal GC poll insertion phase. This only impacts x86.

This change also stops creating inline GC polls for these cases in unoptimized code.

Author:	AndyAyersMS
Assignees:	AndyAyersMS
Labels:	area-CodeGen-coreclr
Milestone:	-

[AndyAyersMS]

@jakobbotsch PTAL
cc @dotnet/jit-contrib

[EgorBo]

Assert failure(PID 29 [0x0000001d], Thread: 217 [0x00d9]): Assertion failed 'doesMethodHaveRecursiveTailcall()' in 'System.Reflection.SignatureType:GetHashCode():int:this' during 'Morph - Global' (IL size 7; hash 0xe30863bc; Instrumented Tier1)

    File: /__w/1/s/src/coreclr/jit/morph.cpp Line: 7459
    Image: /root/helix/work/correlation/dotnet

can it be related? (jitstress) can be also pre-existing issue or my PR

[jakobbotsch]

The assert was added in #93764. I've opened #93844 for it.