Enable tests for consistency of FT sensor and "fake link" frames also to iCub v2 models

[traversaro]

To prevent in the future bugs like #275 .

[traversaro]

Given #275, I was quite surprised that the test pass fine, as there is no l_arm_ft frame in iCub v2 model, so how it is possible that the test compares its value w.r.t. to the pose of the sensor and gets a consistent value? It turns out that indeed the tests is bugged due to a bug in model parsing propagation ad the iDynTree level: robotology/idyntree#1191, so the code in

icub-models-generator/tests/icub-model-test.cpp

Lines 628 to 633 in 5f1654a

	// By default iDynTree creates an additional frame with the same name of the sensor,
	// however in this case we have both the sensor and the urdf frame called <prefix>_ft,
	// and for this test we want to make sure that the <prefix>_ft additional frame is the
	// one in the URDF
	parserOptions.addSensorFramesAsAdditionalFrames = false;
	mdlLoader.setParsingOptions(parserOptions);

was not working as intended, and so it does not fail if the l_arm_ft frame is missing, as the iDynTree URDF parser is automatically creating it out of the sensor pose (and in that case clearly the test will pass, as the frame and sensor pose for sure they will match).

However, this PR already adds a few improvements in testing that I think we can merge as they are, we can open a new PR to update the idyntree version and so see what failures we have once the checkAllFTMeasurementFrameGivenBySensorTagsIsCoherentWithMeasurementFrameGivenByFrame function actually test something.