Add diagnostic code to catch possible space id or page no errors in the log

  Add fault tolerance: when a log record contains nonsensical space id or page number, stop log scan there

added test case that supposedly will crash mysqld built with 2.96

This should cause fewer seeks when using replication.

fixed skip counter to do the RightThing (TM) when the user messes up.
Still do RightThing when he gets it right.

Fixed race condition when using the binary log and INSERT DELAYED which could cause the binary log to have rows that was not yet written to MyISAM tables.

Fixed bug where one got an empty set instead of a DEADLOCK error when using BDB tables.

into mysql.sashanet.com:/home/sasha/src/bk/mysql

on a signal

signal. Tested to the extent that it does not break Linux. Could not test on
Darwin because the machine was not reachable. Will push anyway - at least it
does not make things worse. Testing on Darwin is needed, however.

it uses now thd->lex.ftfunc_list
moved ft-initialization to a separate function
re-disabled implicit ft initialization

  Fix a seg fault if a lock wait timeout is exceeded at the some time someone grants the lock

was up to date.

  Fix wrong (though equivalent) type decl

  Fix a bug in insert buffer B-tree upper levels; probably caused the crash by B.Fitzpatrick
buf0flu.c:
  Fix a bug in previous change
  A small optimization for LRU flushes to avoid losing hot pages from the buffer pool

  Fix a bug which could cause InnoDB to complain it cannot find free blocks from the buffer cache during recovery