Avoid vec_assign() call when all rows match

[DavisVaughan]

For left and full joins, we can often avoid a vec_assign() call if all rows in x have a match in y. This helps performance in these cases.

These benchmarks are on top of #4975

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

data_size <- 1000000

left_df <- tibble(
  a = sample(1:7, data_size, TRUE),
  b = sample(c("a","a","b","c","d"), data_size, TRUE)
)

right_df <- tibble(
  b = c("a","b","c", "d"),
  c = 1:4
)

bench::mark(
  left_join = left_join(left_df, right_df, by = "b"),
  check = FALSE,
  iterations = 100
)

# This PR + #4975 
#> # A tibble: 1 x 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 left_join    46.2ms   50.5ms      19.2    58.5MB     158.

# Just #4975 
#> Warning: Some expressions had a GC in every iteration; so filtering is disabled.
#> # A tibble: 1 x 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 left_join    55.7ms   73.9ms      11.8    73.9MB     23.7

^{Created on 2020-03-11 by the reprex package (v0.3.0)}