However, my impression of the average ETF is that it contains a basket of multiple securities, no?
Not necessarily. There are hundreds of different ETFs and they have very different characteristics. COIN will basically be a tracking ETF, tracking the spot price of a single currency/commodity. A good example for this is the GLD ETF, which tracks the spot price of gold.
If so, why would a single-security ETF (e.g. COIN) be affected by #5 at all?
#5 is true for any security, not just for ETFs. Thinly traded markets are illiquid, easy to manipulate, and cause a lot of slippage (i.e., losses because of the difference between bid and ask).
The effort required to track a single security (e.g. Bitcoin) should be trivial, and therefore very efficient, should it not?
The effort
is trivial - I could probably code it on a PC in a couple of weeks (and most of that time will be spent on getting acquainted with the APIs of the exchanges and on testing). It's not the tracking algorithm that is the problem. It's that if the market is thinly traded (few buyers and sellers), it's easy for an entity to push it in a particular direction with a relatively small amount of money.
Also, you can lose even if the price doesn't move at all. For instance, suppose that there are only 1 buyer and 1 seller besides you and you buy 1 share at $45 from the seller and immediately try to sell it to the buyer, but he is offering only $35. Even if the price didn't move at all (because there were no other participants), you will lose money by exiting your trade.
I've seen both of these things happen in several thinly traded markets, like gold exploration companies, pink sheet shares and so on. It won't necessarily be a problem of COIN, but the advice quoted is a general investment advice, it is not aimed at a particular investment vehicle. We'll have to see how this ETF trades, in order to know if this is going to be a problem.