For your enlightenment, then:
Firstly, “8 per state” is ridiculous from the get-go because it’s not a state-by-state poll. It’s a national poll. 405 from a population is a quite substantial sample (sampling error +/- 5% in this case) for any size population. It’s also pretty much standard for primary polling. For example, the sample size for the latest CNN poll was 444. For the latest CBS/NYT, it was 376. For Monmouth, it was 366. In other words, this is just a regular ol’ poll. How many people do you think pollsters actually poll, anyway? What is your MOE standard to make a poll “a poll”, and do you realize how many people you’d have to call just for that little bit of improvement in confidence?
The 8 per state was clearly rhetorical, and you know that.
As for the sample size of 405 being substantial, it depends on how many choices are being offered. With 15 names on the ballot 405 is nowhere near large enough.