Add tests for not (monadic ~)

[pmikkelsen]

No description provided.

[pmikkelsen]

This pull request adds tests for monadic ~ (not), but it also refactors _RunVariations_ so it runs the model function, instead of the caller. In my opinion, this is the only correct thing to do, as the plan is to add more and more random "variations" of the tests to that operator, which requires that it can run the model and compare the result.

It also means it now becomes possible to pass data to _RunVariations_ which results in errors being generated from the primitive and model function. The errors are compared, just like non-errors are compared.

This pull-request will break all other tests because they use _RunVariations_ the old way, where the right argument contained the expected data as well, often computed as such ... RunVariations (d1 model d2) d1 d2. The fix there is simply to remove (d1 model d2), as it will be computed automatically. For monadic calls such as ... RunVariations (model data) data, the fix is to change it to ... RunVariations ⊂data.

A lot of tests pass explicitly computed "expected values" to _RunVariations_, which is in my opinion better handled by calls to Assert. RunVariations is not able to do anything meaningful with a model function, if the expected values are pre-computed.

[pmikkelsen]

I see that auto-format in RIDE has caused all the changes to become a bit difficult to see. I apologise for that.

[pmikkelsen]

The tests introduced in this pull-request requires the fix for mantis 21762 to be merged into the interpreter.

[sloorush]

Thanks @pmikkelsen

When would the fix for mantis 21762 be merged into the interpreter?