Improve mutex handling when generating stacktraces with NANO_TIMED_LOCKS/txn tracker

[wezrule]

A new CMake variable is introduced, which can be used like so:
-DNANO_TIMED_LOCKS_IGNORE_BLOCKED=ON

Which turns off the timings with NANO_TIMED_LOCKS for mutexes which are blocked while waiting for other thread to release the mutex. This necessarily needs to be done after acquiring the mutex so cannot avoid generating a stacktrace while holding the mutex (which is slow on Windows). With the database transaction tracker it can avoid generating the stacktrace while acquiring the track mutex, so this is now done (TSAN ran with the rpc.database_txn_tracker test).

It does not help in the case where an unrelated mutex has been acquired prior to these operations.

Updated some comments