You largely answered this question yourself: you provided this reference to a Gennero Zezza paper: (link to paper)
Zezza states in the second paragraph that most of the thinking about the Paradox of Profits was mistaken, as the authors used models which were not stock-flow consistent (flows disappeared). If people are arguing that there is a paradox, but do the accounting wrong, it is difficult to explain the error - since their arguments are internally inconsistent.
I am not familiar with Marx’s arguments, but based on the question, he is making an argument about a fixed money supply. This adds an extra constraint, that is probably not addressed in the Zezza papper.
For a single time period, there is no difficulty in generating profits. The business sector starts with \$100, pays \$90 in wages, \$10 in dividends, and then the household spends all money received to buy goods - generating \$10 in profits, which is enough to cover the dividend. Profits can be achieved also by the business sector investing in fixed capital or inventories. In any event, this can be achieved with a fixed money supply.
The only challenge is generating a steady state. Can a fixed money supply co-exist with a growing economy in steady state? This is perhaps a challenge, but the reality is that even non-financial entities can issue instruments that are used as money if the money supply is constrained. (For example, the pre-Confederation Canadian colonies had a perpetual shortage of coinage, and a wide variety of entities issued tokens that circulated as money). Since it is hard for a society to survive without government, government-issued money is normally issued in a way to grow in line with the economy.
Marx was writing in an era before fiat currencies had shown much ability to survive. My understanding is that his thinking did not take into account the flexibility of fiat systems.
In summary, since the argument that profits are paradoxical is based on faulty logic, it’s difficult to explain where the problems lie. One would need to delve into the details of each argument, and determine where they made the error - and it’s not necessarily always the same error.