As a follow-on from SERVER-19981, it would be very useful when trying to track down an unknown source of noTimeout cursors if queries that produced them were logged at the default level.

That is, in addition to the millis > slowms logic at logLevel 0, queries that have QueryOption_NoCursorTimeout set and produce a cursor (ie. not exhausted by the first batch and ntoreturn >= 0), should also be logged by default.