HyphenationCompoundWordTokenFilter returns terms shorter than minSubwordSize [LUCENE-8185]

To account for languages which use binding characters ("fogemorphemes") for composing words the `HyphenationCompoundWordTokenFilter` re-checks the dictionary for a candidate with the last character removed when the original candidate was not found. It currently does not re-check against `minSubWordSize` in this case. Terms that are one character shorter than `minSubWordSize` can be returned.

---
Migrated from [LUCENE-8185](https://issues.apache.org/jira/browse/LUCENE-8185) by Matthias Krueger (@mkr), updated Mar 06 2018
Attachments: [LUCENE-8185.patch](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-8185/LUCENE-8185.patch)
Pull requests: https://github.com/apache/lucene-solr/pull/326

Start with HyphenationCompoundWordTokenFilter and review the attached LUCENE-8185.patch and pull request 326. Check the candidate re-check path described in the issue and add or update a regression test showing that returned terms meet minSubWordSize.

Written by the indexing model from the issue text.