ArabicRootsAnalyzer [LUCENE-4293]

ArabicRootsAnalyzer is using an index of Arabic terms associated with its roots. each Arabic word has a root. There is no automatic way of deciding the root.

This Analyzer will match any term with its root, searching/indexing will be based on roots. It gives me great results in my application.

attached all the required files with the db. the problem with it is the size of the db (16MB). number of terms is around 300,000. I have another db with 600,000 but the attached one is summarized and better i believe.

---
Migrated from [LUCENE-4293](https://issues.apache.org/jira/browse/LUCENE-4293) by Ibrahim
Attachments: [ArabicRootFilter.java](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-4293/ArabicRootFilter.java), [ArabicRootsAnalyzer.java](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-4293/ArabicRootsAnalyzer.java), [ArabicTokenizer.java](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-4293/ArabicTokenizer.java), [ArabicTokens.txt](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-4293/ArabicTokens.txt), [rootsTableIndex.zip](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-4293/rootsTableIndex.zip)

Review the attached ArabicRootFilter.java, ArabicRootsAnalyzer.java, ArabicTokenizer.java, ArabicTokens.txt, and rootsTableIndex.zip first. The issue proposes root-based Arabic searching but does not define an integration point, acceptance criteria, or resolution for the 16MB database, so those requirements need clarification before implementation can be considered done.

Written by the indexing model from the issue text.