As stated by people with long experience here, the limit of the daisy chained 74HC595 is far longer from two, and is related to its electronics and the propagation of the clock signal. After doing my own research some time ago about those displays, the reason most examples and libraries to use two 74HC595 is that one is used for driving the 8 segments of each display digit, the other is connected to the digit enabler, rising the number of digits managed to 8 display ports, 8 led segments each. This might be the cheapest display arrangement in chips, but the resources used to keep the displays updated is the most expensive.
Then you can have a strip of daisy chained 74HC595, each one holding the contents of a single display digit, this will avoid constant refresh, but will force you to send the full content of all the shift registers if a single segment changes.
Finally you have a handfull of display drivers, no only the MAX7219: TM163X series, HT16K33 are probably the most popular.
One final tiny comment: none of those controllers define the colons used as a time units separator as an extra independent element to be set or reset, so you'll have to provide your own logic to drive them, maybe wiring them to the decimal point of some position not using the decimal point, maybe as some segment of an extra digit, in this last case you'll have to count the colons of your display as an extra digit.
@rm-rf - knowing more about what you are up to allows better advices.
You don't have to string all the shift registers or MAX7219s together on one i/o pin. As @gabygold points out, this means changing one item anywhere requires putting out the entire strings of bits.
You could, and probably should, hide that conceptually at a higher level. But a better match at the low level would be to have each logical group of digits (three digit score, N digit this, M digit that) be on its own output pin. Or pins. At least the clock pin can be shared through use of enable.
One thing the shift register won't do is drive huge current for the display. If the MAX7219 can, it would be a good choice. If it cannot, you'll need to hand something off the shift register outputs.
I don't think you've said what Arduino board you are using, but a Mega would provide plenty of pins, or you could challenge yourself and use another shift register (chain maybe) to handle group selection.
All of which means you are back to being able to develop your code with wokwi.
Marking @alto777 answer as correct. I think this thread has gotten off topic. I was specifically looking for a simulation of the MAX7219 chip which doesn't exist, but I can use 74HC595 shift registers which are available in simulators instead.