Fix nullablility rules for field access

[JSAbrahams]

Summary

• Fix bug where incorrect constraints were generated in the unify stage for function arguments in calls

Now, the expression at the site must be the super of any of the types mentioned in the union of types of the argument. We do this by introducing a boolean any in the Name struct. This then changes the behavior when set to true of is_superset_of, where if other's any is true, then self will the super of the other if just one TrueName in the union is a super of the other. Not sure if this logic is completely right. Perhaps it should be that for each TrueName in self, it should be the super of at least one. But I'm leaving it as-is for now.

Added Tests

Happy

• Assign to nullable class field
• Assign to null variable in function body

Sad

• Random test uncommented where we assign to a non-mutable field

[codecov[bot]]

Codecov Report

:exclamation: No coverage uploaded for pull request base (v-0-3-5@986ffda). Click here to learn what that means. The diff coverage is n/a.

@@            Coverage Diff             @@
##             v-0-3-5     #394   +/-   ##
==========================================
  Coverage           ?   87.39%           
==========================================
  Files              ?      109           
  Lines              ?    12028           
  Branches           ?        0           
==========================================
  Hits               ?    10512           
  Misses             ?     1516           
  Partials           ?        0