[SQLite] Add .columnNames property on Cursor

[geelen]

(Replaces #696 as I renamed the remote branch name and the other PR got closed)

This PR adds a .columnNames property that can be called on a Cursor before iterating over the result:

// Given the following data:
sql.exec(`CREATE TABLE abc (a INT, b INT, c INT);`)
sql.exec(`CREATE TABLE cde (c INT, d INT, e INT);`)
sql.exec(`INSERT INTO abc VALUES (1,2,3),(4,5,6);`)
sql.exec(`INSERT INTO cde VALUES (7,8,9),(1,2,3);`)

const iterator = sql.prepare(`SELECT * FROM abc, cde`)();
console.log(iterator.columnNames)
// ["a","b","c","c","d","e"]
console.log(Array.from(iterator.raw()))
// [
//   [1, 2, 3, 7, 8, 9],
//   [1, 2, 3, 1, 2, 3],
//   [4, 5, 6, 7, 8, 9],
//   [4, 5, 6, 1, 2, 3],
// ]

This helps us get around the issue where the default iterator swallows data when column names collide:

const iterator = sql.prepare(`SELECT * FROM abc, cde`)();
console.log(Array.from(iterator))
// [
//   { a: 1, b: 2, c: 7, d: 8, e: 9 },
//   { a: 1, b: 2, c: 1, d: 2, e: 3 },
//   { a: 4, b: 5, c: 7, d: 8, e: 9 },
//   { a: 4, b: 5, c: 1, d: 2, e: 3 },
// ]

As a separate PR, we could make the default iterator error if duplicate columns are detected, but I want to move D1 entirely over to using .all() internally first.

[jasnell]

LGTM but let's have @kentonv sign off also.