ScriptedInitialisationHandler: catch boost exceptions and throw ChimeraTK logic_error with improved what() message

The message of this exception is usually lost and the user is left with
an obscure error about not calling shutdown() in the constructor without
any clue what was wrong with the config

The new order has the following advantages:
- values 0=OK and 1=Fault are compatible with frequent use of booleans like "is_error"
- values are matching ("accidentally") the DOOCS convention

After discussion with Martin K. we concluded that this monitor makes no sense. Legit cases should use the ExactMonitor
instead (if e.g. a boolean on/off configuration switch is to be observed). The monitoring of a Status value makes
no sense, since we would then have a monitor monitoring another monitor. In that case, the behaviour of the other
monitor should either be corrected, or the monitors should be aggregated with the right aggregator mode.

- introduce StatusOutput class
- rename ChimeraTK::States enum to ChimeraTK::StatusOutput::Status and make strongly-typed
- introduce reserved tags to detect StatusOutputs to be aggregated

Before it had to be stripped for all use cases anyway, so it is better to leave it out from the beginning.

Multiplier module: add modern constructor accepting qualified path names of input, factor and output

* add HierarchyModifyingGroup

* VirtualModule: allow nested oneLevelUp etc.; simplify implementation

This is achieved by properly setting the _owner of each VirtualModule (unless highest level).

The simplification comes from then being able to use getOwner() instead of keeping track of "grandparents", and from using createAndGetSubmodule() instead of implementing something similar again in findTagAndAppendToModule().

* DataLossCounter constructor now takes qualified path to trigger variable

* Fix initial msg. Add automatic connection to the LoggingModule.

Now the Logger is a VariableGroup and in the prepare() method an empty
msg is written once. This makes sure the LoggingModule is not waiting
for initial values of Loggers and messages by other Loggers get lost.

* Add message in case on modules with Logger are found.

* Fix test and remove implementation of removed method.

* Do not use hard coded tags but use default tags.

Now all tags given to the LoggingModule are used to look for senders.

The user can now also choose to publish messages only to the control
system.

Refactored.

Reading was done only in deeper levels inside the VariableGroups.

See 1af92147. Input accessors are now properly part of the ReadAny group which improves the deterministic behaviour.

This provides a strict guarantee for the order of updates w.r.t. trigger, especially for tests.

  - Removed extra call to registerModule which caused multiple creation
    of the same module on a tree node
  - Made addChildNode private