JDK 9: TestIndexWriterMerging.testForceMergeDeletes fails with impossible (?) assert trip [LUCENE-6642]

I suspect there is a JVM bug in JDK 9 here... in this code I don't see how current can be null at this point.

Build failure is here: http://build-eu-00.elastic.co/job/lucene_linux_java8_64_test_only/53666/console

```
[junit4] Suite: org.apache.lucene.index.TestIndexWriterMerging
[junit4] 2> NOTE: reproduce with: ant test -Dtestcase=TestIndexWriterMerging -Dtests.method=testForceMergeDeletes -Dtests.seed=D33F5440CCAADDDB -Dtests.slow=true -Dtests.locale=vi -Dtests.timezone=Pacific/Enderbury -Dtests.asserts=true -Dtests.file.encoding=UTF-8
[junit4] FAILURE 0.16s J0 | TestIndexWriterMerging.testForceMergeDeletes <<<
[junit4] > Throwable #1: java.lang.AssertionError
[junit4] > at __randomizedtesting.SeedInfo.seed([D33F5440CCAADDDB:E14F04E6F3205DF4]:0)
[junit4] > at org.apache.lucene.util.fst.BytesStore.writeBytes(BytesStore.java:104)
[junit4] > at org.apache.lucene.store.DataOutput.copyBytes(DataOutput.java:278)
[junit4] > at org.apache.lucene.util.fst.FST.(FST.java:344)
[junit4] > at org.apache.lucene.util.fst.FST.(FST.java:323)
[junit4] > at org.apache.lucene.codecs.blocktreeords.OrdsFieldReader.(OrdsFieldReader.java:83)
[junit4] > at org.apache.lucene.codecs.blocktreeords.OrdsBlockTreeTermsReader.(OrdsBlockTreeTermsReader.java:147)
[junit4] > at org.apache.lucene.codecs.blocktreeords.BlockTreeOrdsPostingsFormat.fieldsProducer(BlockTreeOrdsPostingsFormat.java:93)
[junit4] > at org.apache.lucene.codecs.perfield.PerFieldPostingsFormat$FieldsReader.(PerFieldPostingsFormat.java:261)
[junit4] > at org.apache.lucene.codecs.perfield.PerFieldPostingsFormat.fieldsProducer(PerFieldPostingsFormat.java:341)
[junit4] > at org.apache.lucene.index.SegmentCoreReaders.(SegmentCoreReaders.java:104)
[junit4] > at org.apache.lucene.index.SegmentReader.(SegmentReader.java:65)
[junit4] > at org.apache.lucene.index.StandardDirectoryReader$1.doBody(StandardDirectoryReader.java:58)
[junit4] > at org.apache.lucene.index.StandardDirectoryReader$1.doBody(StandardDirectoryReader.java:50)
[junit4] > at org.apache.lucene.index.SegmentInfos$FindSegmentsFile.run(SegmentInfos.java:731)
[junit4] > at org.apache.lucene.index.StandardDirectoryReader.open(StandardDirectoryReader.java:73)
[junit4] > at org.apache.lucene.index.DirectoryReader.open(DirectoryReader.java:63)
[junit4] > at org.apache.lucene.index.TestIndexWriterMerging.testForceMergeDeletes(TestIndexWriterMerging.java:151)
[junit4] > at java.lang.Thread.run(Thread.java:745)
[junit4] 2> NOTE: test params are: codec=Asserting(Lucene53): {tvtest=FST50, count=PostingsFormat(name=Memory doPackFST= false), id=PostingsFormat(name=Memory doPackFST= false), termVector=BlockTreeOrds(blocksize=128)}, docValues:{}, sim=RandomSimilarityProvider(queryNorm=true,coord=yes): {tvtest=DFR I(n)B3(800.0), id=DFR I(n)L2, termVector=DFR GLZ(0.3)}, locale=vi, timezone=Pacific/Enderbury
[junit4] 2> NOTE: Linux 3.13.0-52-generic amd64/Oracle Corporation 1.9.0-ea (64-bit)/cpus=8,threads=1,free=132149080,total=307232768
[junit4] 2> NOTE: All tests run in this JVM: [TestBytesRef, TestDateSort, TestTermsEnum2, TestIOUtils, TestDocBoost, TestIndexWriterConfig, TestTimeLimitingCollector, TestMultiValuedNumericRangeQuery, TestBagOfPostings, TestConsistentFieldNumbers, TestFieldCacheRewriteMethod, TestFieldType, TestDuelingCodecs, TestSloppyMath, TestMathUtil, TestReusableStringReader, TestSloppyPhraseQuery2, TestTerm, TestExitableDirectoryReader, TestSizeBoundedForceMerge, TestInPlaceMergeSorter, TestLongPostings, TestTermRangeFilter, TestCachingCollector, TestSameTokenSamePosition, TestIndexWriterMerging]
[junit4] Completed [167/402] on J0 in 10.39s, 6 tests, 1 failure <<< FAILURES!
```

---
Migrated from [LUCENE-6642](https://issues.apache.org/jira/browse/LUCENE-6642) by Michael McCandless (@mikemccand)

Reproduce TestIndexWriterMerging.testForceMergeDeletes with the ant command and seed shown in the issue. Trace the failure through BytesStore.writeBytes, DataOutput.copyBytes, FST, and the blocktreeords readers, then determine the cause of the assertion failure and whether the test passes after the issue is addressed.

Written by the indexing model from the issue text.