[Fix #47] add_reference_concurrently not checking for mismatching types

[joshuay03]

Fixes #47

The issue was that add_reference_concurrently did not have its commands checked. I've now sent it through the same checker for add_reference while ensuring the bad index check always passes.

[fatkodima]

Thank you! ❤️

Let me know if you need a new release.

[ghiculescu]

@fatkodima sort of related to this, do you have any views on making check_mismatched_foreign_key_type work in one direction only? For example:

• You have a "customers" table that has as an int ID column.
• You are creating a new table with an FK to "customers". You make customers_id a bigint.
• At the moment check_mismatched_foreign_key_type will raise an error.

Unless there's some postgres behaviour I'm not aware of that would break it, I think this behaviour should be supported. Assuming you know that customers has the wrong ID type (if this gem doesn't already, we should think of ways to alert you about that), there's no reason to force into using the wrong column type in other places as well.

The inverse case (customers is bigint; your new FK is int) should not be allowed.

[fatkodima]

I think the gem can just implicitly ignore the cases when we add a bigint foreign key referencing the integer primary key.
I will implement this in a couple of next days or feel free to send a PR, if you would like.

[joshuay03]

I think the gem can just implicitly ignore the cases when we add a bigint foreign key referencing the integer primary key. I will implement this in a couple of next days or feel free to send a PR, if you would like.

Got this up: #49