Don't try this when working on the FreeDB. Again,
too much hassle to unkink the recursions...

Only loosen src page if it's actually dirty.
Refactor page refs in page_merge.

Pages that were dirtied and deleted in the same txn should be
reused, instead of consuming freeDB pages.

Similar to prev commit for mdb_node_move, the call of
mdb_page_serch_lowest() was leaving csrc unusable.

Was leaving stale info in dst cursor

Use num*20, chosen from empirical testing

Was skipping the overwrite in some cases, if the cmp function
said the new and old already matched. We should always perform
the overwrite anyway, since the cmp function may only be doing
a prefix compare and the data may actually be different.

Due to commit 18a07eb7

With mdb_cursor_put(new key, data item big enough to need a sub-DB):
- Track cursors as when adding other nodes.  The "!do_sub" is now
  unneeded.  Before ITS#7861 it distinguished inserting key from data.
- Set MDB_TXN_ERROR if an empty sub-DB was created.  I.e. adding
  key:subDB succeeded, but inserting the data item failed.

Update the state before looping, so the key is not duplicated.
Broke in 4c8f5761 (mostly).

mdb_cursor_del(): Fix md_entries with MDB_NODUPDATA.
mdb_cursor_put(): Check in xcursor whether an item got inserted. Count
data instead of keys after 13d6d59c.

Since non-NULL key must be provided now.

Shift the rest of the cursor stack as needed

Make sure to propagate rebalance results all the
way back to original caller's cursor.

Fixup cursor state after node_move/page_merge

Fix md_entries count, test for insert on md_entries increment

Note MDB_CURRENT/MDB_DUPSORT requirement