GH-46600: [C++][CI] Add job with ARROW_LARGE_MEMORY_TESTS enabled (#49490)

### Rationale for this change

Now that we have self-hosted runners with AWS we should test the `ARROW_LARGE_MEMORY_TESTS` on CI.

### What changes are included in this PR?

Added new runner for ARROW_LARGE_MEMORY_TESTS.
Fix `parquet-writer-test` to generate huge expected page using huge `max_rows_per_page` instead of default.

### Are these changes tested?

Yes via CI

### Are there any user-facing changes?

No

* GitHub Issue: #46600

Authored-by: Raúl Cumplido <[email protected]>
Signed-off-by: Raúl Cumplido <[email protected]>

Author: raulcd

.github/workflows/cpp_extra.yml

@@ -107,6 +107,16 @@ jobs:
           - image: alpine-linux-cpp
             runs-on: ubuntu-latest
             title: AMD64 Alpine Linux
+          - image: ubuntu-cpp
+            run-options: >-
+              -e ARROW_CTEST_TIMEOUT=2000
+              -e ARROW_C_FLAGS_DEBUG="-O1"
+              -e ARROW_CXX_FLAGS_DEBUG="-O1"
+              -e ARROW_GANDIVA=OFF
+              -e ARROW_LARGE_MEMORY_TESTS=ON
+              -e BUILD_WARNING_LEVEL=PRODUCTION
+            runs-on: "runs-on=${{ github.run_id }}/family=x8i.2xlarge/volume=80gb/spot=capacity-optimized"
+            title: AMD64 Ubuntu Large Memory Tests
           - image: conda-cpp
             run-options: >-
               -e ARROW_USE_MESON=ON

cpp/src/parquet/column_writer_test.cc

@@ -1053,8 +1053,10 @@ TEST(TestColumnWriter, LARGE_MEMORY_TEST(WriteLargeDictEncodedPage)) {
                       {
                           PrimitiveNode::Make("item", Repetition::REQUIRED, Type::INT32),
                       }));
-  auto properties =
-      WriterProperties::Builder().data_pagesize(1024 * 1024 * 1024)->build();
+  auto properties = WriterProperties::Builder()
+                        .data_pagesize(1024 * 1024 * 1024)
+                        ->max_rows_per_page(std::numeric_limits<int64_t>::max())
+                        ->build();
   auto file_writer = ParquetFileWriter::Open(sink, schema, properties);
   auto rg_writer = file_writer->AppendRowGroup();
 
@@ -1124,8 +1126,10 @@ TEST(TestColumnWriter, LARGE_MEMORY_TEST(ThrowsOnDictIndicesTooLarge)) {
                       {
                           PrimitiveNode::Make("item", Repetition::REQUIRED, Type::INT32),
                       }));
-  auto properties =
-      WriterProperties::Builder().data_pagesize(4 * 1024LL * 1024 * 1024)->build();
+  auto properties = WriterProperties::Builder()
+                        .data_pagesize(4 * 1024LL * 1024 * 1024)
+                        ->max_rows_per_page(std::numeric_limits<int64_t>::max())
+                        ->build();
   auto file_writer = ParquetFileWriter::Open(sink, schema, properties);
   auto rg_writer = file_writer->AppendRowGroup();