Apply[Read].ap reads the columns in the wrong order

[nigredo-tori]

Currently Apply[Read].ap is implemented like this:

def ap[A, B](ff: Read[A => B])(fa: Read[A]): Read[B] = fa.ap(ff)

That is, it first reads fa, and then ff. This leads to surprising results when manually building a Read instance:

scala> val read = (Read[Int], Read[String]).tupled
read: doobie.util.Read[(Int, String)] = doobie.util.Read@4fb0497f

scala> val query = sql"select 1, 'foo'".query(read)
query: doobie.Query0[(Int, String)] = doobie.util.query$Query$$anon$3@419a4cd0

scala> query.check.unsafeRunSync
  Query0[(Int, String)] defined at <console>:50
  select 1, 'foo'
  ✓ SQL Compiles and TypeChecks
  ✕ C01 ?column? INTEGER (int4) NULL?  →  String
    INTEGER (int4) is ostensibly coercible to String according to the
    JDBC specification but is not a recommended target type. Expected
    schema type was CHAR or VARCHAR or LONGVARCHAR or NCHAR or
    NVARCHAR or LONGNVARCHAR.
  ✕ C02 ?column? VARCHAR (text) NULL?  →  Int
    VARCHAR (text) is ostensibly coercible to Int according to the
    JDBC specification but is not a recommended target type. Expected
    schema type was INTEGER.

[nigredo-tori]

Discussion on a similar topic: typelevel/cats#2172