fix: JsonTreeWriter accepts finite BigDecimal/BigInteger like JsonWriter

The tree writer's finiteness check used doubleValue(), which can overflow
to Infinity for finite BigDecimal/BigInteger values like 1E400 —
causing toJsonTree to reject inputs that toJson accepts.

JsonWriter.value(Number) already special-cases the
'alwaysCreatesValidJsonNumber' classes (BigDecimal, BigInteger,
AtomicInteger, AtomicLong) to skip the doubleValue-based oracle. This
change mirrors that guard in JsonTreeWriter.value(Number), making the
two writer surfaces agree.

Existing tests (ToNumberPolicyTest, JsonTreeWriterTest, etc.) continue
to pass; the new JsonTreeWriterFiniteBigDecimalTest now passes.

Author: andrewstellman

gson/src/main/java/com/google/gson/internal/bind/JsonTreeWriter.java

@@ -25,6 +25,8 @@
 import com.google.gson.stream.JsonWriter;
 import java.io.IOException;
 import java.io.Writer;
+import java.math.BigDecimal;
+import java.math.BigInteger;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.Objects;
@@ -218,7 +220,10 @@ public JsonWriter value(Number value) throws IOException {
       return nullValue();
     }
 
-    if (!isLenient()) {
+    // BigDecimal/BigInteger are always valid JSON numbers (mirrors
+    // JsonWriter.alwaysCreatesValidJsonNumber); their doubleValue() can overflow to Infinity
+    // even though the value is finite, so do not use it as the finiteness oracle for them.
+    if (!isLenient() && !(value instanceof BigDecimal) && !(value instanceof BigInteger)) {
       double d = value.doubleValue();
       if (Double.isNaN(d) || Double.isInfinite(d)) {
         throw new IllegalArgumentException("JSON forbids NaN and infinities: " + value);