Constraints (initially CHECK and FOREIGN KEY) should be represented in pg-roll's internal schema representation.

Also, UNIQUE constraints should be part of the internal schema representation too.

Having constraints represented in the internal schema would enable:

• Better validation of operations involving constraints:
  • Ensure that a constraint of the same name does not exist when adding a new constraint.
  • Ensure that the constraint exists when dropping a constraint.
• Not having to specify the name of the column on which the constraint is defined when dropping a constraint (and the confusion that occurs when the wrong column is specified)

### Tasks
- [x] Validate constraint exists before dropping (https://github.com/xataio/pgroll/pull/252)
- [x] Ensure constraint of same name doesn't exist when adding an `FK` constraint
- [x] Ensure constraint of same name doesn't exist when adding a `CHECK` constraint
- [x] Ensure constraint of same name doesn't exist when adding a `UNIQUE` constraint
- [x] BREAKING CHANGE: no longer require `Column` field when dropping a constraint