Windows: Use FrontendPtr for extra TLS space on WOW64/ARM64EC

We have run out of EmulatorData slots on ARM64EC, so introduce this for
less-frequently accessed TLS data for which the additional cost doesn't
matter.

Source/Windows/WOW64/Module.cpp

@@ -107,6 +107,10 @@ struct TLS {
   }
 };
 
+struct FrontendThreadData {
+  bool InLockedRWXRead {};
+};
+
 class WowSyscallHandler;
 
 namespace {
@@ -142,6 +146,10 @@ TLS GetTLS() {
   return TLS {NtCurrentTeb()};
 }
 
+FrontendThreadData* GetFrontendThreadData(FEXCore::Core::InternalThreadState* Thread) {
+  return static_cast<FrontendThreadData*>(Thread->FrontendPtr);
+}
+
 uint64_t GetWowTEB(void* TEB) {
   static constexpr size_t WowTEBOffsetMemberOffset {0x180c};
   return static_cast<uint64_t>(
@@ -595,6 +603,8 @@ void BTCpuThreadInit() {
   TLS.ThreadState() = Thread;
   TLS.ControlWord().fetch_or(ControlBits::WOW_CPU_AREA_DIRTY, std::memory_order::relaxed);
 
+  Thread->FrontendPtr = new FrontendThreadData();
+
   auto ThreadTID = GetCurrentThreadId();
   Threads.emplace(ThreadTID, Thread);
   if (StatAllocHandler) {
@@ -641,6 +651,8 @@ void BTCpuThreadTerm(HANDLE Thread, LONG ExitCode) {
   }
   auto ThreadState = TLS.ThreadState();
 
+  delete GetFrontendThreadData(ThreadState);
+
   // GDT and LDT are mirrored, only free one.
   delete[] ThreadState->CurrentFrame->State.segment_arrays[FEXCore::Core::CPUState::SEGMENT_ARRAY_INDEX_GDT];