My opinion: Arbitration earnings can be obtained by large institutional investors. They have access to information of places where the price of an asset is different from that another place, and with large amounts of money can take advantage of that price differential, while it lasts, since demand and supply will cancel the advantage of arbitration and the prices will be the same again.
Yes, large insititutional investors or full-time individual investor with knowledge will be able to identify underprice or overprice securities using the model and may be able to make abnormal profits.
Yes, as aforementioned, in real life you take advantage of the misspricing after identify there is a under-over valuation of the stock. Arbitrage is someway positive by bringing efficiency to the price of the stock, or valuation throughout the markets. As an investor, you take advantage by getting the differential profit of the valuation until reach its intrinsic value.
Yes my understanding is that there are two things. one is the arbitrage opportunity in the spot market where the parity conditions are not satisfied and the same thing is traded at different markets and investors those who are alert can capitalize. The other is the arbitrage opportunity in the investment market where the expected return is not equal to the factor values and factor sensitivities.In other words when the asset is over/under priced which is arrived at on your analysis and can be used to make investment portfolios.
There is a decent article in Wikipedia on APT. Based on Wikipedia arbitrage in APT is when you sell short "overpriced" assets and buy "underpriced" one . Keep in mind that "overpriced"/"underpriced" assets are in term of APT model, not in reality. So it is not a real arbitrage.
There is book "Active Portfolio Management" by Grinold, Kahn. Where second author is portfolio manager based on some form of APT.
"Investment Analysis" by F. Reilly once was in CFA course - very nice and include APT chapter
All good observations, to which I would add that a critical issue in arbitrage is its persistence. Do abnormal opportunities have an unlimited shelf life? In my view, this is rarely the case, and most arbs will admit that they need to "tune" their systems from time to time and hunt constantly for new opportunities. In my paper https://ssrn.com/abstract=2925532, I offer a lifecycle model for the birth and eventual death of strategies, which in particular applies to arbitrage strategies. For a specific instance of the "death" of a life cycle, see Khandani & Lo, "What Happened to the Quants in August, 2007."