Fix BufferManager::FindOrCreate Recreation Bugs

It was determined that `FindOrCreate` has several issues which this commit fixes:
* It wouldn't correctly handle locking of the newly created `Buffer` as the constructor would setup traps prior to being able to lock it which could lead to UB
* It wouldn't propagate the `usedByContext`/`everHadInlineUpdate` flags correctly
* It wouldn't correctly set the `dirtyState` of the buffer according to that of its source buffers

app/src/main/cpp/skyline/gpu/buffer.h

@@ -140,13 +140,6 @@ namespace skyline::gpu {
 
         Buffer(GPU &gpu, GuestBuffer guest);
 
-        /**
-         * @brief Creates a Buffer that is pre-synchronised with the contents of the input buffers
-         * @param tag The tag to associate locking of srcBuffers with
-         * @param srcBuffers Span of overlapping source buffers
-         */
-        Buffer(GPU &gpu, GuestBuffer guest, ContextTag tag, span<std::shared_ptr<Buffer>> srcBuffers);
-
         /**
          * @brief Creates a host-only Buffer which isn't backed by any guest buffer
          * @note The created buffer won't have a mirror so any operations cannot depend on a mirror existing
@@ -222,6 +215,13 @@ namespace skyline::gpu {
          */
         bool PollFence();
 
+        /**
+         * @brief Invalidates the Buffer on the guest and deletes the trap that backs this buffer as it is no longer necessary
+         * @note This will not clear any views or delegates on the buffer, it will only remove guest mappings and delete the trap
+         * @note The buffer **must** be locked prior to calling this
+         */
+        void Invalidate();
+
         /**
          * @brief Synchronizes the host buffer with the guest
          * @param rwTrap If true, the guest buffer will be read/write trapped rather than only being write trapped which is more efficient than calling MarkGpuDirty directly after