This generates random numbers from a zero-truncated Poisson distribution,
i.e. from `X | X > 0`

when `X ~ Poisson(lambda)`

. The advantage here is that
we guarantee to return exactly `n`

numbers and without using a loop internally.
This solution was provided in a post by
Peter Dalgaard.

## Arguments

- n
(`numeric`

)

Number of random numbers.

- lambda
(`numeric`

)

Non-negative mean(s).

## Value

The random numbers.

## Examples

```
x <- rpois(1e6, lambda = 5)
x <- x[x > 0]
hist(x)
y <- rtpois(1e6, lambda = 5)
hist(y)
```