org.apache.lucene.analysis.tests.TestRandomChains.testRandomChains test failure

### Description

Haven't finished bisection to figure out the origin of the bug.

```
org.apache.lucene.analysis.tests.TestRandomChains > testRandomChains FAILED
java.lang.IllegalStateException: last stage: inconsistent endOffset at pos=1: 8 vs 11; token=ڎbdb觸 Ǧ
at __randomizedtesting.SeedInfo.seed([3AEBBB09EA70DCE:3E4F92D1D9B5100E]:0)
at org.apache.lucene.test_framework@10.0.0-SNAPSHOT/org.apache.lucene.tests.analysis.ValidatingTokenFilter.incrementToken(ValidatingTokenFilter.java:164)
at org.apache.lucene.test_framework@10.0.0-SNAPSHOT/org.apache.lucene.tests.analysis.BaseTokenStreamTestCase.checkAnalysisConsistency(BaseTokenStreamTestCase.java:1289)
at org.apache.lucene.test_framework@10.0.0-SNAPSHOT/org.apache.lucene.tests.analysis.BaseTokenStreamTestCase.checkRandomData(BaseTokenStreamTestCase.java:1187)
at org.apache.lucene.test_framework@10.0.0-SNAPSHOT/org.apache.lucene.tests.analysis.BaseTokenStreamTestCase.checkRandomData(BaseTokenStreamTestCase.java:1081)
at org.apache.lucene.analysis.tests@10.0.0-SNAPSHOT/org.apache.lucene.analysis.tests.TestRandomChains.testRandomChains(TestRandomChains.java:945)
at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:103)
```

```
2> stage 0: ڎ<[1-2] +1> d<[4-5] +1> 觸<[7-8] +1> Ǧ<[10-11] +1>
2> stage 1: ڎ<[1-2] +1> d<[4-5] +1> 觸<[7-8] +1> Ǧ<[10-11] +1>
2> stage 2: ڎ<[1-2] +1> ڎ d<[1-5] +0> d<[4-5] +1> d 觸<[4-8] +0> 觸<[7-8] +1> 觸 Ǧ<[7-11] +0>
2> last stage: ڎbdb觸<[1-8] +1> ڎbdb觸 Ǧ<[1-11] +0>
2> TEST FAIL: useCharFilter=true text='\u0003\ufb87\udacd\uddf6d\uf1f4\u02e6\u89f8\uda06\udfd2\u01e6'
2> Exception from random analyzer:
2> charfilters=
2> org.apache.lucene.analysis.icu.ICUNormalizer2CharFilter(java.io.StringReader@24c09603, com.ibm.icu.impl.Norm2AllModes$ComposeNormalizer2@43b7067e)
2> tokenizer=
2> org.apache.lucene.analysis.th.ThaiTokenizer(org.apache.lucene.util.AttributeFactory$1@a0c154ba)
2> filters=
2> org.apache.lucene.analysis.no.NorwegianMinimalStemFilter(ValidatingTokenFilter@215f24e9 term=,bytes=[],startOffset=0,endOffset=0,positionIncrement=1,positionLength=1,type=word,termFrequency=1,flags=0,payload=null,keyword=false)
2> org.apache.lucene.analysis.shingle.ShingleFilter(ValidatingTokenFilter@47f9add2 term=,bytes=[],startOffset=0,endOffset=0,positionIncrement=1,positionLength=1,type=word,termFrequency=1,flags=0,payload=null,keyword=false, fg)
2> org.apache.lucene.analysis.shingle.FixedShingleFilter(ValidatingTokenFilter@71cd79a term=,bytes=[],startOffset=0,endOffset=0,positionIncrement=1,positionLength=1,type=word,termFrequency=1,flags=0,payload=null,keyword=false, 3, b, )
> java.lang.IllegalStateException: last stage: inconsistent endOffset at pos=1: 8 vs 11; token=ڎbdb觸 Ǧ
```

### Gradle command to reproduce

```
./gradlew test --tests TestRandomChains.testRandomChains -Dtests.seed=3AEBBB09EA70DCE -Dtests.nightly=true -Dtests.locale=en-CA -Dtests.timezone=America/Cordoba -Dtests.asserts=true -Dtests.file.encoding=UTF-8
```

Run the supplied Gradle command for TestRandomChains.testRandomChains with the recorded seed. Start at TestRandomChains.java:945 and inspect ValidatingTokenFilter.java:164, following the reported ThaiTokenizer, ShingleFilter, and FixedShingleFilter chain. Done means the inconsistent endOffset failure is understood and the reproducing test passes.

Written by the indexing model from the issue text.