Add new Lint/AmbiguousAssignment cop

[fatkodima]

Closes #9223

[dvandersluis]

Maybe AmbiguousAssignment would be a better name?

[dvandersluis]

Also, I think we need to ensure that x = -y (for example) is allowed, since it'll parse to the same AST as x =- y, and is valid. The offense message should probably suggest that x =- y could be either x -= y or x = -y since both are valid and we can't know what they were going for.

[bbatsov]

I think that's why there's no auto-correct for the cop - it's hard to tell if that's a bug or poorly formatted code. That being said it seems to me the code simply checks in the source if the operator is right after the =, so it wouldn't result in a false positive for x = -y. That being said, some tests to cover this and +/*y would be useful.

[fatkodima]

For me both names (current and AmbiguousAssignment) are good enough, not sure which is better.
Expanded tests.

[dvandersluis]

AmbiguousAssignment would be parallel with the other Lint cops (AmbiguousOperator, AmbiguousBlockAssociation, AmbiguousRegexpLiteral), but I'd leave it up to @bbatsov

[bbatsov]

Let's go with Ambiguous. Consistent naming is a good thing.

[fatkodima]

Changed the name.

[bbatsov]

Thanks!