build(simdjson): Add CMake option to skip utf8 validation (#12165)

Summary:
Add VELOX_SIMDJSON_SKIPUTF8VALIDATION CMake option to do this.

Discussed in #10639 (comment)

Pull Request resolved: #12165

Reviewed By: kgpai

Differential Revision: D68754000

Pulled By: xiaoxmeng

fbshipit-source-id: 4a5e43dd1c14a17c52d896521af30699b1b541a6

Author: marin-ma

CMake/resolve_dependency_modules/simdjson.cmake

@@ -29,4 +29,8 @@ FetchContent_Declare(
   URL ${VELOX_SIMDJSON_SOURCE_URL}
   URL_HASH ${VELOX_SIMDJSON_BUILD_SHA256_CHECKSUM})
 
+if(${VELOX_SIMDJSON_SKIPUTF8VALIDATION})
+  set(SIMDJSON_SKIPUTF8VALIDATION ON)
+endif()
+
 FetchContent_MakeAvailable(simdjson)

CMakeLists.txt

@@ -149,6 +149,8 @@ option(
   VELOX_ENABLE_INT64_BUILD_PARTITION_BOUND
   "make buildPartitionBounds_ a vector int64 instead of int32 to avoid integer overflow when the hashtable has billions of records"
   OFF)
+option(VELOX_SIMDJSON_SKIPUTF8VALIDATION
+       "Skip simdjson utf8 validation in JSON parsing" OFF)
 
 # Explicitly force compilers to generate colored output. Compilers usually do
 # this by default if they detect the output is a terminal, but this assumption