[NativeAOT-LLVM] Move RhpPInvoke and undefine BACKGROUND_GC

[SingleAccretion]

See also: #2472.

As somewhat expected, undefining BACKGROUND_GC lets us drop a cool 56K of dead native code:

Summary of Code Size diffs:
(Lower is better)

Total bytes of base: 3249451
Total bytes of diff: 3193293
Total bytes of delta: -56158 (-1.73% % of base)
Average relative delta: -6.87%
    diff is an improvement
    average relative diff is an improvement

Top method regressions (percentages):
        1311 (323.70% of base) : 1130.dasm - WKS::AllocAlign8(alloc_context*, WKS::gc_heap*, unsigned long, unsigned int)
         634 (139.34% of base) : 1131.dasm - WKS::GCHeap::Alloc(gc_alloc_context*, unsigned long, unsigned int)
         768 (50.53% of base) : 1048.dasm - WKS::gc_heap::copy_brick_card_table()
         210 (40.94% of base) : 1133.dasm - WKS::GCHeap::GetTotalBytesInUse()
         124 (32.46% of base) : 1047.dasm - WKS::gc_heap::make_card_table(unsigned char*, unsigned char*)
         135 (25.42% of base) : 1051.dasm - WKS::gc_heap::delete_heap_segment(WKS::heap_segment*, int)
           3 ( 9.38% of base) : 1009.dasm - RhpPInvoke
          42 ( 5.27% of base) : 1158.dasm - S_P_TypeLoader_Internal_Runtime_TypeLoader_GenericDictionaryCell_MethodLdTokenCell__Create

Top methods only present in diff:
         410 (     ∞ of base) : 1162.dasm - WKS::gc_heap::pm_full_gc_init_or_clear()

Top method improvements (percentages):
        -208 (-99.05% of base) : 1140.dasm - WKS::GCHeap::PublishObject(unsigned char*)
         -72 (-97.30% of base) : 1012.dasm - WKS::GCHeap::WaitUntilConcurrentGCComplete()
         -61 (-93.85% of base) : 1141.dasm - WKS::GCHeap::WaitUntilConcurrentGCCompleteAsync(int)
         -11 (-84.62% of base) : 1143.dasm - WKS::GCHeap::TemporaryDisableConcurrentGC()
         -11 (-84.62% of base) : 1142.dasm - WKS::GCHeap::TemporaryEnableConcurrentGC()
         -21 (-84.00% of base) : 1144.dasm - WKS::GCHeap::IsConcurrentGCEnabled()
         -10 (-71.43% of base) : 1014.dasm - WKS::GCHeap::IsConcurrentGCInProgress()
       -1627 (-62.24% of base) : 1070.dasm - WKS::gc_heap::a_fit_free_list_uoh_p(unsigned long, alloc_context*, unsigned int, int, int)
        -799 (-60.21% of base) : 1060.dasm - WKS::gc_heap::init_semi_shared()
         -16 (-59.26% of base) : 1129.dasm - WKS::GCHeap::GetPromotedBytes(int)
         -67 (-55.83% of base) : 1137.dasm - WKS::GCHeap::SetGcLatencyMode(int)
        -146 (-49.16% of base) : 1083.dasm - WKS::gc_heap::get_total_survived_size()
         -40 (-48.78% of base) : 1135.dasm - WKS::GCHeap::CollectionCount(int, int)
        -642 (-45.47% of base) : 1081.dasm - WKS::gc_heap::joined_generation_to_condemn(int, int, int, int*)
         -85 (-44.74% of base) : 1128.dasm - WKS::GCHeap::IsPromoted(Object*)
        -926 (-41.64% of base) : 1074.dasm - WKS::gc_heap::garbage_collect(int)
        -351 (-40.11% of base) : 1080.dasm - WKS::gc_heap::try_allocate_more_space(alloc_context*, unsigned long, unsigned int, int)
        -695 (-38.85% of base) : 1040.dasm - WKS::gc_heap::grow_brick_card_tables(unsigned char*, unsigned char*, unsigned long, WKS::heap_segment*, WKS::gc_heap*, int)
        -303 (-36.55% of base) : 1132.dasm - WKS::GCHeap::GarbageCollect(int, bool, int)
        -850 (-35.52% of base) : 1079.dasm - WKS::gc_heap::allocate_uoh(int, unsigned long, alloc_context*, unsigned int, int)

Top methods only present in base:
         -37 (-100.00% of base) : 1000.dasm - CLREventStatic::CreateAutoEventNoThrow(bool)
       -1749 (-100.00% of base) : 1122.dasm - WKS::gc_heap::background_process_mark_overflow_internal(unsigned char*, unsigned char*, int)
         -86 (-100.00% of base) : 1078.dasm - WKS::gc_heap::trigger_gc_for_alloc(int, gc_reason, WKS::GCDebugSpinLock*, bool, WKS::msl_take_state)
        -767 (-100.00% of base) : 1125.dasm - WKS::gc_heap::bgc_thread_function()
       -1651 (-100.00% of base) : 1126.dasm - WKS::gc_heap::revisit_written_page(unsigned char*, unsigned char*, int, unsigned char*&, unsigned char*&, int, unsigned long&)
        -789 (-100.00% of base) : 1127.dasm - WKS::gc_heap::bgc_tuning::set_total_gen_sizes(bool, bool)
        -264 (-100.00% of base) : 1134.dasm - WKS::GCHeap::ApproxTotalBytesInUse(int)
        -240 (-100.00% of base) : 1072.dasm - WKS::gc_heap::trigger_ephemeral_gc(gc_reason, WKS::enter_msl_status*)
        -635 (-100.00% of base) : 1069.dasm - WKS::exclusive_sync::uoh_alloc_set(unsigned char*)
        -418 (-100.00% of base) : 1068.dasm - WKS::gc_heap::bgc_uoh_alloc_clr(unsigned char*, unsigned long, alloc_context*, unsigned int, int, int, int, int, WKS::heap_segment*)
        -612 (-100.00% of base) : 1121.dasm - WKS::gc_heap::scan_background_roots(void (*)(Object**, ScanContext*, unsigned int), int, ScanContext*)
        -737 (-100.00% of base) : 1064.dasm - WKS::exclusive_sync::init()
        -192 (-100.00% of base) : 1061.dasm - WKS::gc_heap::create_bgc_threads_support(int)
          -2 (-100.00% of base) : 1057.dasm - WKS::gc_heap::fire_alloc_wait_event_end(WKS::alloc_wait_reason)
          -2 (-100.00% of base) : 1056.dasm - WKS::gc_heap::fire_alloc_wait_event_begin(WKS::alloc_wait_reason)
        -826 (-100.00% of base) : 1055.dasm - WKS::gc_heap::reset_write_watch(int)
        -248 (-100.00% of base) : 1054.dasm - WKS::gc_heap::background_delay_delete_uoh_segments()
       -1560 (-100.00% of base) : 1106.dasm - WKS::gc_heap::bgc_tuning::calculate_tuning(int, bool)
        -216 (-100.00% of base) : 1052.dasm - WKS::gc_heap::decommit_heap_segment(WKS::heap_segment*)
        -958 (-100.00% of base) : 1147.dasm - SoftwareWriteWatch::GetDirty(void*, unsigned long, void**, unsigned long*, bool, bool)

[SingleAccretion]

@dotnet/nativeaot-llvm

[yowl]

LGTM!