Add a mock/test codec that redirects all doc values formats to BINARY [LUCENE-9205]

We have some exciting proposed changes that will use `BINARY` doc values, e.g. [using block compression in the default codec](https://github.com/apache/lucene-solr/pull/1234#pullrequestreview-353073188), supporting approximate nearest-neighbor search using [HNSW](https://issues.apache.org/jira/browse/LUCENE-9004) or [IVFFLAT](https://issues.apache.org/jira/browse/LUCENE-9136), but I think our testing of `BINARY` doc values could be improved.

A possibly simple way to do that would be to make a mock/test codec that wraps the default codec but implements all doc values formats (`NUMERIC`, `SORTED`, etc.) on top of the wrapped codec's `BINARY` format. This way the wrapped codec only ever sees `BINARY`, and then all Lucene tests that test the other doc values codecs, would be testing `BINARY`.

---
Migrated from [LUCENE-9205](https://issues.apache.org/jira/browse/LUCENE-9205) by Michael McCandless (@mikemccand)

Start by tracing the default codec and its NUMERIC, SORTED, and BINARY doc values formats. Determine how a mock/test codec could wrap the default codec while routing every format through BINARY, then verify that the existing Lucene tests exercise the other doc values types through that path.

Written by the indexing model from the issue text.