Remove support for the no-cache configuration, revert back to maintaining only a pair of hazard pointers at most (a parent/child pair), rather than the full stack of hazard pointers (one per level of the tree).

Alex, I don't think there's anything here you need to review, I just pulled "no_cache" out of the LSM cursor create call and everything else was pretty obvious.

Michael, you might want to look at my new routine *wt_bt_swap/*wt_page_swap_func in btree.i, and, of course, where I'm using it in bt_walk.c, row_srch.c and col_srch.c, I think the rest if pretty innocent.

I tucked in one other change in this branch, removing the split-percentage and overflow-size checking whenever WT_COMPRESSOR::compress_raw is configured, this should move compress_raw toward a true page-layout function, as we discussed.