* fixed catastrophic typo causing bogus count-query results from branch server
* disabled "vacuous get_hit_strings request" check introduced in v2.0.31
- unable to reproduce any noticeable turnaround delay using bz, bz_pp, or bz_meta on kira
- allowing vacuous 'get_hit_strings' requests was in fact often faster than without (but only by ca. 1ms in fs-cached case)
- dominant factor for query time is fs-cache presence of get_hit_strings *results* (doesn't apply to empty get_hit_strings requests)
+ fscache @ ca. 10 ms/q ; -fscache @ 430 ms/q
- for v2.1.10, we need get_hit_strings even for "empty" daughter result-sets in order to update navHint
* disabled trimming of CFirstHitsQueryResult::m_Hits[] in BranchServer.cpp
- we're paying for the memory anyways, and we need all retrieved hits for consistent navHints in some cases
* fixed inconsistent navigation hint generation if all hits were trimmed (requires 'vacuous' get_hit_strings requests)