Go: set subtypes column to true for models where it has a meaning

[owen-mc]

This is almost always the behaviour that we want. I will do a QA run to see what alert changes this leads to.

[smowton]

I'd say we might as well leave this set to False where we're modelling a function, not a method, if only to avoid confusing a reader wondering what a "subtype" means in that context

[owen-mc]

I was thinking it was easier to have the convention that we always set it to True unless there was a very good reason to, so we don't have to think about what is a function and what is a method. Admittedly it's pretty obvious from whether the "type" column is empty or not, but I still think some people will set it to False for methods by mistake at some point if we mix True and False.

[owen-mc]

QA showed 3 extra alerts. I was able to reproduce one locally and it was a valid result. I wasn't able to reproduce the other two locally. I don't think it's worth chasing it down too much - the small number of extra alerts reassures me that this PR is okay to merge.

[mbg]

Regarding the discussion about what to set this to for functions, my view is:

• Ideally, there would be no subtype column at all for models where that has no meaning or doesn't make sense. As is the case for functions here.
• But since MaD doesn't allow that kind of distinction (AFAIK), it may be easier to have a non-boolean value here that indicates that it's not applicable (e.g. unused or the empty string).

[owen-mc]

@mbg 's higher-minded suggestions notwithstanding, more people seem to be in favour of having subtypes=False for rows where it makes no sense (which is precisely when the types column is empty). I have pushed a commit making the change. For the record there are 553 rows with the types column empty and 827 rows with the types column non-empty.

[smowton]

Looks sensible if eval results are positive.