fix: return-statement reachability analysis

[anton-trunov]

Closes #268

• I have updated CHANGELOG.md
• I have added unit tests to demonstrate the contribution is correctly implemented
• I have run all the tests locally and no test failure was reported
• I did not do unrelated and/or undiscussed refactorings

[novusnota]

How could we fix the stmts-failed/case-41.tact in the future? Recursive descent of sorts?

Other than that, splendid work!

[anton-trunov]

Thanks @novusnota!

How could we fix the stmts-failed/case-41.tact in the future? Recursive descent of sorts?

It can be done for the simple case of Booleans, but not in general.

Consider the following example:

if belongs(x, aSetOfIntegers) {
    return 1;
}
if belongs(x, anotherSetOfIntegers) {
    return 2;
}

if aSetOfIntegers \union anotherSetOfIntegers == Int then we have covered all the paths, but it's too expensive to try and solve it (or even undecidable). Btw, Rust solves this issue in the same way essentially, but it has if-expressions so it's a typechecking error.

You can try using SMT solvers (or your own custom solver) to do this, but it's not worth it in practice.