Use thrust::identity as hash functions for byte pair encoding (#13665)

This PR fixes a minor issue that distinct hash functions are used for `insert` and `find` in byte pair encoding. It also verifies that the latest changes in cuco won't break cudf.

Authors:
  - Yunsong Wang (https://github.com/PointKernel)

Approvers:
  - David Wendt (https://github.com/davidwendt)
  - https://github.com/nvdbaranec
  - Mark Harris (https://github.com/harrism)

URL: #13665

cpp/src/text/subword/bpe_tokenizer.cu

@@ -38,6 +38,7 @@
 #include <thrust/execution_policy.h>
 #include <thrust/find.h>
 #include <thrust/for_each.h>
+#include <thrust/functional.h>
 #include <thrust/iterator/counting_iterator.h>
 #include <thrust/iterator/transform_iterator.h>
 #include <thrust/merge.h>
@@ -234,7 +235,7 @@ struct byte_pair_encoding_fn {
         if (rhs.empty()) break;  // no more adjacent pairs
 
         auto const hash    = compute_hash(lhs, rhs);
-        auto const map_itr = d_map.find(hash);
+        auto const map_itr = d_map.find(hash, thrust::identity<cudf::hash_value_type>{});
         if (map_itr != d_map.end()) {
           // found a match; record the rank (and other min_ vars)
           auto const rank = static_cast<cudf::size_type>(map_itr->second);

cpp/src/text/subword/load_merges_file.cu

@@ -119,7 +119,7 @@ std::unique_ptr<detail::merge_pairs_map_type> initialize_merge_pairs_map(
 
   merge_pairs_map->insert(iter,
                           iter + input.size(),
-                          cuco::murmurhash3_32<cudf::hash_value_type>{},
+                          thrust::identity<cudf::hash_value_type>{},
                           thrust::equal_to<cudf::hash_value_type>{},
                           stream.value());