[muh] add tuning_rle_encode.cuh: BI-V100 tuning for rle_encode

Translated from CCCL with SMEM overflow protection.
All SM100 values checked against 48KB limit.

muh/include/muh/tuning/tuning_rle_encode.cuh

@@ -0,0 +1,54 @@
+// muh/include/muh/tuning/tuning_rle_encode.cuh — BI-V100
+//
+// Mirrors: cccl_upstream/cub/cub/device/dispatch/tuning/tuning_rle_encode.cuh
+// CCCL SM100: 14 type specializations, all tiles ≤ 28672 (safe for 48KB)
+//
+// vllm relevance: attention mask compression via run-length encoding
+
+#pragma once
+
+#include "muh/hardware.cuh"
+#include "muh/tuning/common.cuh"
+
+namespace muh::tuning::rle_encode {
+
+struct RleEncodePolicy {
+  int threads_per_block;
+  int items_per_thread;
+  BlockLoadAlgorithm load_algorithm;
+  CacheLoadModifier load_modifier;
+  BlockScanAlgorithm scan_algorithm;
+  LookbackDelayPolicy delay;
+};
+
+struct policy_selector {
+  int item_size;
+  int length_size;
+
+  constexpr RleEncodePolicy operator()(const hardware_capability& hw) const {
+    // SM100 patterns: threads=192-448, items=7-15
+    // All tiles ≤ 28672, no overflow risk on BI-V100
+    int threads = 256;
+    int items = 10;
+    
+    // Scale items by type size (larger types → fewer items)
+    if (item_size >= 8) {
+      items = 7;
+    } else if (item_size >= 4) {
+      items = 10;
+    } else {
+      items = 14;
+    }
+    
+    // SMEM check: tile = threads * items * (item_size + length_size)
+    int pair_size = item_size + length_size;
+    while (threads * items * pair_size > hw.max_shared_memory_per_block && items > 1)
+      items--;
+
+    return {threads, items, BLOCK_LOAD_WARP_TRANSPOSE, LOAD_DEFAULT,
+            BLOCK_SCAN_WARP_SCANS,
+            {LookbackDelayAlgorithm::fixed_delay, 350, 450}};
+  }
+};
+
+} // namespace muh::tuning::rle_encode