toString() methods on term/queries/etc are wrong: assume utf-8 encoded bytes. [LUCENE-2942]

In Lucene's trunk, a Term is just a Bytesref.
In a lot of cases this is a UTF-8 encoded string, but in some cases its not (e.g. collation fields).

The problem is that the toString methods all currently call utf8ToString().
This is wrong, though from a practical point of view i think just printing the bytes won't be very helpful for debugging most cases where the bytes really are utf-8 encoded.

So i think in these cases we should use the following technique: if the bytes are a valid utf-8 sequence, use BytesRef.utf8tostring(), otherwise just print the bytes: BytesRef.toString()

its no problem for performance because toString is only for debugging anyway.

---
Migrated from [LUCENE-2942](https://issues.apache.org/jira/browse/LUCENE-2942) by Robert Muir (@rmuir)
Attachments: [LUCENE-2942.patch](https://apache.github.io/lucene-jira-archive/attachments/LUCENE-2942/LUCENE-2942.patch)

Start by locating the Term, query, and related toString() methods that call BytesRef.utf8ToString(), then read BytesRef.utf8ToString() and BytesRef.toString(). Done means valid UTF-8 bytes retain readable output while invalid sequences use the byte representation, with the affected toString() methods covered by the relevant tests.

Written by the indexing model from the issue text.