Phrase slop ignored after any nested parser in same query

_query_:"{!term f=text}foo" "foo bar"~1

is parsed as

text:foo "foo bar"

ignoring phrase slop because subQParser field in SolrQueryParserBase is not cleared after magic field processing.

Likely introduced in SOLR-4625 (May 2013) and may be the double-slop query mentioned in prior issue should throw an exception, not silently dropping additional slop, but it will be a breaking change because it is very-very old parser trait.