Write tests that only valid recovery accessors are written to the device

[killenb]

Child of #129

Only recovery accessors which have seen their initial value must be written to the device. Prepare a test scenario for this.

There are two devices. Device A is not functional yet (use exception dummy and make it throw on open), Device B is working.

• Have an ApplicationModule which has a poll-type input from Device A. Don't connect it to the CS so no fan is involved.

• It has a second input from the CS, which is push type (a trigger)

• The only output is directly connected to a variable in Device B.

• The device module main loop first assigns the value of the poll-type input to the output, then writes to the output. Afterwards it waits for the trigger, then polls the other input and cycles the loop.

• Before starting the app, prepare the target register in Device B such that it is not 0. Prepare the source register in Device A such that it has a different value than the register in B, also not 0.

• Check that the value in B is never overwritten with 0

• After removing the exception state of Device A the initial value is propagated without having to send a trigger.

• If you now change the value in A and send the trigger, the new values is propagated as usual.

Extended Scenario: You could have a second register written to B from the CS. This works even if A is still disfunctional. If you now break B, send another value from the CS, then 'repair' B, only the second value from the CS is recovered. The one that is supposed to come from A is still not touched, especially not overwritten with 0, when recovering B.

The test will only succeed if #144 is implemented.

[killenb]

I took away the umbrella child because the parent was an implementation ticket, this one is about tests.
It is part of #180

[mhier]

Now on redmine #8368