fix Spark RDD生成成功，但是dump失败问题

最近碰到一个很奇葩的问题，RDD成功产生（rdd.take(100).mkstring("\n")成功打出），但是dump到hdfs失败，报如下的错误

org.apache.parquet.io.InvalidRecordException: could not get child 3 from [GroupColumnIO user r:0 d:1 [user], GroupColumnIO items r:0 d:1 [items], PrimitiveColumnIO expId r:0 d:1 [expId]]
	at org.apache.parquet.io.GroupColumnIO.getChild(GroupColumnIO.java:114)
	at org.apache.parquet.thrift.ParquetWriteProtocol$StructWriteProtocol.(ParquetWriteProtocol.java:322)
	at org.apache.parquet.thrift.ParquetWriteProtocol$MessageWriteProtocol.(ParquetWriteProtocol.java:397)
	at org.apache.parquet.thrift.ParquetWriteProtocol.(ParquetWriteProtocol.java:431)
	at org.apache.parquet.hadoop.thrift.AbstractThriftWriteSupport.prepareForWrite(AbstractThriftWriteSupport.java:121)
	at org.apache.parquet.hadoop.t

java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
	at java.lang.Thread.run(Thread.java:748)
Caused by: java.lang.IndexOutOfBoundsException: Index: 3, Size: 3
	at java.util.ArrayList.rangeCheck(ArrayList.java:657)
	at java.util.ArrayList.get(ArrayList.java:433)
	at org.apache.parquet.io.GroupColumnIO.getChild(GroupColumnIO.java:112)
	... 19 more

RDD的类型定义定义

struct MCRecommendSamplesBatch {
    1: optional MCRankerUserInfo.MCRankerUserInfo user;
    2: optional MCRankerUserInfo.RankerContextInfo context; // 暂时没用到
    3: optional list<RecommendImpression> items;
    4: optional string expId;

}

 看错误的显示就是没找到context，这个太扯淡了，搞了半天发现是 RankerContextInfo 定义的结构体为空导致的，加了个字段就没事了。

struct RankerContextInfo {  // 原来因为没用到context特征，啥也没定义，导致报错

}

struct RankerContextInfo { // 改成这样就可以了
	1:optional string debug;	

}