[SPARK-13790] Speed up ColumnVector's getDecimal

## What changes were proposed in this pull request?

We should reuse an object similar to the other non-primitive type getters. For
a query that computes averages over decimal columns, this shows a 10% speedup
on overall query times.

## How was this patch tested?

Existing tests and this benchmark

```
TPCDS Snappy:                       Best/Avg Time(ms)    Rate(M/s)   Per Row(ns)
--------------------------------------------------------------------------------
q27-agg (master)                       10627 / 11057         10.8          92.3
q27-agg (this patch)                     9722 / 9832         11.8          84.4
```

Author: Nong Li <nong@databricks.com>

Closes #11624 from nongli/spark-13790.

sql/catalyst/src/main/java/org/apache/spark/sql/catalyst/expressions/UnsafeRow.java

@@ -400,7 +400,7 @@ public final class UnsafeRow extends MutableRow implements Externalizable, KryoS
       return null;
     }
     if (precision <= Decimal.MAX_LONG_DIGITS()) {
-      return Decimal.apply(getLong(ordinal), precision, scale);
+      return Decimal.createUnsafe(getLong(ordinal), precision, scale);
     } else {
       byte[] bytes = getBinary(ordinal);
       BigInteger bigInteger = new BigInteger(bytes);

sql/catalyst/src/main/scala/org/apache/spark/sql/types/Decimal.scala

@@ -376,6 +376,17 @@ object Decimal {
 
   def apply(value: String): Decimal = new Decimal().set(BigDecimal(value))
 
+  /**
+   * Creates a decimal from unscaled, precision and scale without checking the bounds.
+   */
+  def createUnsafe(unscaled: Long, precision: Int, scale: Int): Decimal = {
+    val dec = new Decimal()
+    dec.longVal = unscaled
+    dec._precision = precision
+    dec._scale = scale
+    dec
+  }
+
   // Evidence parameters for Decimal considered either as Fractional or Integral. We provide two
   // parameters inheriting from a common trait since both traits define mkNumericOps.
   // See scala.math's Numeric.scala for examples for Scala's built-in types.

sql/core/src/main/java/org/apache/spark/sql/execution/vectorized/ColumnVector.java

@@ -536,9 +536,9 @@ public abstract class ColumnVector {
    */
   public final Decimal getDecimal(int rowId, int precision, int scale) {
     if (precision <= Decimal.MAX_INT_DIGITS()) {
-      return Decimal.apply(getInt(rowId), precision, scale);
+      return Decimal.createUnsafe(getInt(rowId), precision, scale);
     } else if (precision <= Decimal.MAX_LONG_DIGITS()) {
-      return Decimal.apply(getLong(rowId), precision, scale);
+      return Decimal.createUnsafe(getLong(rowId), precision, scale);
     } else {
       // TODO: best perf?
       byte[] bytes = getBinary(rowId);