The demand for dollars is internal to the US. Both businesses and consumers pay for things with expected future cash flow — for example, they pay their rent with their income (whether from customers or from a job). Or they took out loans with the expectation that they could pay the monthly payment with that cash flow.

When that cash flow stops — people stop eating out at restaurants, or consumers lose their jobs — they need cash (dollars) to pay their obligations. The supply of money is the same — the same number of dollars exist — but the demand is higher now.

Essentially, everyone is getting margin called all at once. If there is no intervention, businesses and people that are in sustainable & healthy in the long-term go bankrupt and fire workers, and our society loses a lot of organizational capital.

We don't want a short term shock (a pandemic) to hurt our long term growth. To avoid that, the Fed injects some money so that the supply can meet the demand. Then businesses can borrow money to pay their employees (and their rent), plus we don't fall into a deflationary spiral.

Google the term "liquidity preference". It's generally understood to be a function of real GDP and interest rates. GDP up = more demand for currency to be able to transact. Lower rates = holding cash has less opportunity cost (there are fewer opportunities to invest in other things with higher yield).

This appears to be a theoretical term. Is there any empirical chart that demonstrates this, or more generally, "money demand"?

There is no non theoretical way to measure demand and supply separately from market data. All prices, for example, already include not only costs, but expectations of demand.

You need data that includes so called demand shifters or demand multipliers, to estimate the supply and vice versa.

That is, if you observe a supply independent shift of demand, you can use that measure of to identify supply parameters. Even if you di this non parametrically, you implicitly impose a model.

Supply and demand is one of the origin problems of the theory of statistical identification and causal analysis.