Refactor app services and stabilize refresh/UI flows

- Add refresh scheduler and network refresh service boundaries for typed
  refresh results, ordered RPC collectors, applicators, and price parsing.
- Add daemon lifecycle and wallet security workflow helpers while preserving
  App-owned command RPC, decrypt, cancellation, and UI handoff behavior.
- Split balance, console, mining, amount formatting, and async task logic into
  focused modules with expanded Phase 4 test coverage.
- Fix market price loading by triggering price refresh immediately, avoiding
  queue-pressure drops, tracking loading/error state, and adding translations.
- Polish send, explorer, peers, settings, theme/schema, and related tab UI.
- Replace checked-in generated language headers with build-generated resources.
- Document the cleanup audit, UI static-state guidance, and architecture updates.

src/ui/windows/mining_benchmark.h

@@ -0,0 +1,66 @@
+#pragma once
+
+#include <vector>
+
+namespace dragonx {
+namespace ui {
+
+struct ThreadBenchmark {
+    enum class Phase { Idle, Starting, WarmingUp, Measuring, Advancing, CoolingDown, Done };
+
+    struct Result {
+        int threads = 0;
+        double hashrate = 0.0;
+    };
+
+    static constexpr float MIN_WARMUP_SECS = 90.0f;
+    static constexpr float MAX_WARMUP_SECS = 300.0f;
+    static constexpr float MEASURE_SECS = 30.0f;
+    static constexpr float COOLDOWN_SECS = 5.0f;
+    static constexpr float STABILITY_WINDOW_SECS = 10.0f;
+    static constexpr float STABILITY_THRESHOLD = 0.05f;
+    static constexpr int STABLE_WINDOWS_NEEDED = 3;
+
+    Phase phase = Phase::Idle;
+    std::vector<int> candidates;
+    int current_index = 0;
+    std::vector<Result> results;
+    float phase_timer = 0.0f;
+    double prev_window_avg = 0.0;
+    double window_sum = 0.0;
+    int window_samples = 0;
+    float window_timer = 0.0f;
+    int consecutive_stable = 0;
+    double measure_sum = 0.0;
+    int measure_samples = 0;
+    int optimal_threads = 0;
+    double optimal_hashrate = 0.0;
+    bool was_pool_running = false;
+    int prev_threads = 0;
+    float total_warmup_secs = 0.0f;
+
+    void reset();
+    void buildCandidates(int maxThreads);
+    float avgWarmupSecs() const;
+    float perTestSecs() const;
+    float totalEstimatedSecs() const;
+    float elapsedSecs() const;
+    float progress() const;
+    void resetStabilityTracking();
+    bool active() const;
+};
+
+struct ThreadBenchmarkUpdate {
+    bool stopPoolMining = false;
+    bool startPoolMining = false;
+    int startThreads = 0;
+    bool saveOptimalThreads = false;
+    int optimalThreads = 0;
+};
+
+ThreadBenchmarkUpdate AdvanceThreadBenchmark(ThreadBenchmark& benchmark,
+                                             float deltaSeconds,
+                                             double poolHashrate10s);
+
+} // namespace ui
+} // namespace dragonx