Shorter SQL for verb() + distinct()

[mgirlich]

Closes #879.

library(dplyr, warn.conflicts = FALSE)
library(dbplyr, warn.conflicts = FALSE)

lf <- lazy_frame(x = 1, y = 1)

Renamed variable

lf %>%
    select(a = x, y) %>%
    distinct(a)

Before

SELECT DISTINCT `a`
FROM (
  SELECT `x` AS `a`, `y`
  FROM `df`
) `q01`

After

SELECT DISTINCT `x` AS `a`
FROM `df`

Unnecessary mutated variable

lf %>%
    mutate(z = 1) %>%
    group_by(x) %>%
    distinct(y)

Before

SELECT DISTINCT `x`, `y`
FROM (
  SELECT `x`, `y`, 1.0 AS `z`
  FROM `df`
) `q01`

After

SELECT DISTINCT *
FROM `df`

[hadley]

This looks great!

(With the caveat that I'm terrible at coming up with adversarial edge cases that will break the generated SQL)