[MUH] Delete 20 dead-code batch-generated tuning headers

Audit results:
  - 20/20 files had IDENTICAL if-branch and fallback (dead code)
  - 787 lines total, 5% coverage of 15116 lines in CCCL originals
  - No type specializations, no offset_size branches, no benchmark data
  - 0 of 20 algorithms appear on vllm's Qwen3.6 inference hot path

The 6 headers that remain (reduce, topk, scan, transform, batch_memcpy, for)
are the only algorithms that execute during vllm decode/prefill/cache operations.
These 6 have real type specializations and CCCL SM100 reference values.

CCCL has 26 algorithms because it's a general-purpose library.
muh targets one workload: Qwen3.6-35B-A3B on 4× BI-V100.
Covering algorithms that don't execute is worse than not covering them —
it creates the illusion of completeness.

muh/include/muh/tuning/tuning_merge.cuh

@@ -1,39 +0,0 @@
-// muh/include/muh/tuning/tuning_merge.cuh — BI-V100 merge tuning
-//
-// Mirrors: cccl_upstream/cub/cub/device/dispatch/tuning/tuning_merge.cuh
-// vllm impact: sequence merge in batch scheduling
-// Competition weight: Output TPS × 16.796
-
-#pragma once
-
-#include "muh/hardware.cuh"
-#include "muh/tuning/common.cuh"
-
-namespace muh::tuning::merge {
-
-struct MergePolicy {
-  int threads_per_block;
-  int items_per_thread;
-  CacheLoadModifier load_modifier;
-};
-
-struct bi100_default {
-  static constexpr int threads = 512;
-  static constexpr int items = 15;
-  static constexpr int load_mod = LOAD_DEFAULT;
-};
-
-struct policy_selector {
-  int key_size;
-  int value_size;
-
-  constexpr MergePolicy operator()(const hardware_capability& hw) const {
-    if (hw.at_least(hardware_capability::vendor_t::iluvatar, 100)) {
-      return {bi100_default::threads, bi100_default::items, LOAD_DEFAULT};
-    }
-    // Fallback
-    return {bi100_default::threads, bi100_default::items, LOAD_DEFAULT};
-  }
-};
-
-} // namespace muh::tuning::merge