nvflinger: Use a std::array for the available displays instead of std::vector

The built-in set of displays is fixed, so we can utilize an array
instead of a vector here.

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -30,12 +30,6 @@ constexpr std::size_t SCREEN_REFRESH_RATE = 60;
 constexpr u64 frame_ticks = static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
 
 NVFlinger::NVFlinger() {
-    // Add the different displays to the list of displays.
-    displays.emplace_back(0, "Default");
-    displays.emplace_back(1, "External");
-    displays.emplace_back(2, "Edid");
-    displays.emplace_back(3, "Internal");
-
     // Schedule the screen composition events
     composition_event =
         CoreTiming::RegisterEvent("ScreenComposition", [this](u64 userdata, int cycles_late) {

src/core/hle/service/nvflinger/nvflinger.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 #include <string>
 #include <string_view>
@@ -84,7 +85,12 @@ private:
 
     std::shared_ptr<Nvidia::Module> nvdrv;
 
-    std::vector<Display> displays;
+    std::array<Display, 4> displays{{
+        {0, "Default"},
+        {1, "External"},
+        {2, "Edid"},
+        {3, "Internal"},
+    }};
     std::vector<std::shared_ptr<BufferQueue>> buffer_queues;
 
     /// Id to use for the next layer that is created, this counter is shared among all displays.