fix: Avoid dumping unique_constraint when attisdropped

[BuonOmo]

Fixes #347

[mmalek-sa]

This still doesn't handle the indexes with where clauses, for example:

ActiveRecord::Schema.define do
  create_table :payments, force: true do |t|
    t.text 'name'
    t.timestamp 'deleted_at'
    t.index 'lower(name::STRING) ASC', name: 'lower_name_index_on_payments', unique: true
    t.index ['name'], name: 'index_payments_on_name', unique: true, where: '(deleted_at IS NULL)'
  end
end

Will result in a schema dump that lacks deleted_at IS NULL:

ActiveRecord::Schema[7.2].define(version: 0) do
  create_table "payments", force: :cascade do |t|
    t.text "name"
    t.datetime "deleted_at", precision: nil
    t.index "lower(name) ASC", name: "lower_name_index_on_payments", unique: true
    t.unique_constraint ["name"], name: "index_payments_on_name"
  end
end

[mmalek-sa]

This should be:

Suggested change

	UniqueConstraintDefinition.new(table_name, columns, options)
	ActiveRecord::ConnectionAdapters::PostgreSQL::UniqueConstraintDefinition.new(table_name, columns, options)

[BuonOmo]

Columns with the UNIQUE constraint automatically have an index created with the name <table name>_<columns>_key. To avoid having two identical indexes, you should not create indexes that exactly match the UNIQUE constraint's columns and order.

_{(https://www.cockroachlabs.com/docs/stable/unique)}

I think we can just consider every unique_constraint as an index then.

The UNIQUE constraint depends on the automatically created index, so dropping the index also drops the UNIQUE constraint. Conversely, dropping the UNIQUE constraint also drops the automatically created index.

[rafiss]

thanks for this fix!