If the return channel of a FeedingFanOut was removed because it is
connected to the control system but not mapped there, the _returnSlave
is not set and hence all read operations on it were crashing.

* Disable deprecation warnings
* Fix linter findings
* fix: Compiler warnings regarding internal linkage
* fix: Mask memory leak

---------

Co-authored-by: Martin Hierholzer <martin.hierholzer@desy.de>

Rearranged processing of fanouts and their consuming nodes. A list of
consuming nodes and implementations is passed to the fanout constructors,
so the nodes don't have to added later. This resolves conflicts in the
order in which accessors are decorated and added to the fanouts.

FanOut constructors could still use refactoring.

Ongoing refactoring. Consuming fanout works again.

ConstantAccessor not pushes the value to the queue yet

This partly reverts d5ab1447, where the existing doReadTransfer overrides
were removed. This commit gets the existing implmentations back, but
changes the function name to doReadTransferSynchronously.

Remove functions that do not exist anymore

transform logic_errors in doXxxTransferYyy that are already checked in the preXxx functions into asserts

Fixes #153

Fixes #151

Replaces calls to doXxxTransferYyy of the TransferElement by the
corresponding xxxTransferYyy call

- Exception handling decorator and ThreadedFanOut send condition variable notificatios via the DeviceModule
- FanOut sends terminate to impl and all slaves.
Solves (some?) shutdown issues, but breaks bi-directional array test.