Fix value of test requiring integer to use integer

[pasi-paasiala]

This file requires measure to be IfcInteger and simpleValue is a decimal number. I think we should consider this kind of IDS fail illegal. It makes no sense to put extra effort to the implementations to cover for these kind of cases.

[aothms]

This has been discussed here: #89 I think the sentiment there was that (requiring) a 3rd state (fail, pass, invalid IDS) would result in the most additional effort. So requiring failure on this rule is the quickest way towards consistent behaviour among all tools. Validation of the IDS itself can be a later step and is a potentially rather nuanced.

[CBenghi]

Now we have the third state, given the effort put in the auditing of IDSs.
The hope is that this would simplify development effort in the end.

[aothms]

Maybe, the more relevant point to discuss is this file, which passes:

https://github.com/buildingSMART/IDS/blob/master/Documentation/testcases/property/pass-integer_values_are_checked_using_type_casting_2_4.ids

[CBenghi]

Regardless from the decision we take on third state we need to include https://github.com/buildingSMART/IDS/blob/master/Documentation/testcases/property/pass-integer_values_are_checked_using_type_casting_3_4.ids when considering what is a valid value for IDS

[berlotti]

I'd support these states: fail, pass, and an optional 'invalid IDS' (as specialisation of fail).
Implementers are not expected to cast.
Especially with the https://github.com/buildingSMART/IDS-Audit-tool we can support help implementers to check validity of IDS.