Add benchmark tests for CAST INTEGER/BIGINT to DECIMAL (facebookincub…

…ator#9568)

Summary:
This PR adds `CAST INTEGER/BITINT to SHORT/LONG DECIMAL` benchmarking test.

```
============================================================================
[...]hmarks/ExpressionBenchmarkBuilder.cpp     relative  time/iter   iters/s
============================================================================
cast##try_cast_invalid_empty_input                        888.87us     1.13K
cast##tryexpr_cast_invalid_empty_input                    366.60ms      2.73
cast##try_cast_invalid_nan                                482.08ms      2.07
cast##tryexpr_cast_invalid_nan                            834.13ms      1.20
cast##try_cast_valid                                        4.52ms    221.42
cast##tryexpr_cast_valid                                    4.52ms    221.16
cast##cast_valid                                            4.52ms    221.42
cast##cast_decimal_to_inline_string                         2.21ms    451.69
cast##cast_short_decimal                                    3.14ms    318.79
cast##cast_long_decimal                                     5.05ms    197.88
cast##cast_large_real_to_scientific_notation               10.45ms     95.74
cast##cast_small_real_to_standard_notation                  5.77ms    173.40
cast##cast_small_double_to_scientific_notation             12.60ms     79.39
cast##cast_large_double_to_standard_notation                7.34ms    136.21
cast##cast_timestamp                                        3.94ms    253.68
cast##cast_real_as_int                                    404.66us     2.47K
cast##cast_decimal_as_bigint                              969.12us     1.03K
cast##cast_int_as_short_decimal                           911.29us     1.10K
cast##cast_int_as_long_decimal                            925.29us     1.08K
cast##cast_bigint_as_short_decimal                        906.91us     1.10K
cast##cast_bigint_as_long_decimal                         922.87us     1.08K

```

The positive relative time implies Velox is slower compared to DuckDb:
Here is how the relative time computed:

```
 trialResults[actualTrials] = std::make_pair(
          max(0.0, double(nsecs.count()) / timeIterData.niter - globalBaseline),
          std::move(timeIterData.userCounters));
```

Pull Request resolved: facebookincubator#9568

Reviewed By: Yuhta

Differential Revision: D62599114

Pulled By: kevinwilfong

fbshipit-source-id: 1bad04fd2418d2daa84cb45564ae9dd7f68e5818

velox/benchmarks/basic/CastBenchmark.cpp

@@ -48,6 +48,14 @@ int main(int argc, char** argv) {
       vectorSize, [](auto /*row*/) { return "infinity"; });
   auto spaceInput = vectorMaker.flatVector<std::string>(
       vectorSize, [](auto /*row*/) { return "   "; });
+  auto integerInput = vectorMaker.flatVector<int32_t>(
+      vectorSize, [&](auto j) { return 12345 * j; }, nullptr);
+  auto bigintInput = vectorMaker.flatVector<int64_t>(
+      vectorSize,
+      [&](auto j) {
+        return facebook::velox::HugeInt::build(12345 * j, 56789 * j + 12345);
+      },
+      nullptr);
   auto decimalInput = vectorMaker.flatVector<int64_t>(
       vectorSize, [&](auto j) { return 12345 * j; }, nullptr, DECIMAL(9, 2));
   auto shortDecimalInput = vectorMaker.flatVector<int64_t>(
@@ -148,6 +156,8 @@ int main(int argc, char** argv) {
               {"valid",
                "empty",
                "invalid",
+               "integer",
+               "bigint",
                "decimal",
                "short_decimal",
                "long_decimal",
@@ -158,6 +168,8 @@ int main(int argc, char** argv) {
               {validInput,
                emptyInput,
                invalidInput,
+               integerInput,
+               bigintInput,
                decimalInput,
                shortDecimalInput,
                longDecimalInput,
@@ -191,6 +203,14 @@ int main(int argc, char** argv) {
           "cast(large_double as varchar)")
       .addExpression("cast_real_as_int", "cast (small_real as integer)")
       .addExpression("cast_decimal_as_bigint", "cast (short_decimal as bigint)")
+      .addExpression(
+          "cast_int_as_short_decimal", "cast (integer as decimal(18,6))")
+      .addExpression(
+          "cast_int_as_long_decimal", "cast (integer as decimal(38,16))")
+      .addExpression(
+          "cast_bigint_as_short_decimal", "cast (bigint as decimal(18,6))")
+      .addExpression(
+          "cast_bigint_as_long_decimal", "cast (bigint as decimal(38,16))")
       .withIterations(100)
       .disableTesting();