+ fixed bug with extrema-specification in #greater_by_date[LO,HI]
+ problem is more general with descending sorts
- assumedly #greater_by_size, #greater_by_rank, and #greater_by[FIELD] are also affected
- descending sorts SHOULD just wrap ascending variants with a specialied comparison operator,
BUT that wreaks havoc with extrema-based filtering in CConcHolder::GetAllHits()
- underlying weakness seems to be confusion wrt what's being sorted (hits vs hit-indices vs ...),
over-use of literal < operator on CHit.m_OrderId (e.g. in my_lower_bound()), and general wonkiness
wrt sorting vs bounds-checking