Clarify `fmi3GetIntervalDecimal` behaviour depending on `qualifier` value

[nl78]

fmi3GetIntervalDecimal uses the intervals array to return interval values and the qualifiers array to specify how each interval should be interpreted.
The specification could be clarified: when the qualifier fmi3IntervalUnchanged is returned, the corresponding entries in the intervals array are not set and must be ignored.

The same applies to fmi3IntervalNotYetKnown, although this behavior is more intuitive in that case.

[chrbertsch]

The specification could be clarified: when the qualifier fmi3IntervalUnchanged is returned, the corresponding entries in the intervals array are not set and must be ignored.

The spec currently states: "fmi3IntervalUnchanged is returned if a previous call to fmi3GetInterval already returned a value qualified with fmi3IntervalChanged which has not changed since. In Scheduled Execution this means the corresponding model partition has already been scheduled."

For me it is clear that if for qualifier entries with fmi3IntervalUnchanged the corresponding entries in the intervals array are changed since the last call anyway, this would clearly be a buggy implementation. I do not think that we have to clarify that in the buggy case the values have to be ignored ...

[nl78]

I encountered an FMU that returns a FMU which returns fmi3IntervalUnchanged and leaves the corresponding interval value in an undefined state (i.e., the FMU does not update the interval array for this particular clock). I guess this FMU is not buggy.
In this case, the FMI importer should reuse the interval value from the previous call (and may adjust it to account for any time increment).
Initially, I assumed that the interval array would be updated with the already known value—that is, the value from the previous call, possibly adjusted if a time increment had occurred.

[chrbertsch]

Ah, ok, now I anderstand. Perhaps a non-normative "NOTE:" is sufficient here?
@nl78 : Could you please provide a PR with the concrete wording that we could review?

[chrbertsch]

related to #2018