how successful such a solution be in the long run with a growing number of devices?
That's part long-answered (there's libraries full of research into cooperative, non-cooperative, and cognitive medium access control), and partly an ongoing topic of research.
Generally, in these bands, there's strict legal requirements that your device cannot simply occupy the channel for as long as it wants. It needs to limit itself in time, or try to find out whether someone else tries to use the channel, or both.
Also, power control inherently becomes a must in dense network: you're no longer SNR-limited, your interference (SIR) limited, so "shouting louder" makes your life worse (just as at a party, there's a point where it becomes to loud to talk).
So, there's not only approaches, but also laws for solving this at smaller scale problems. This breaks down when coverage areas get large – I'm almost certain LoRa won't be sustainable as wide-area covering system in urban areas, simply because it's woefully underequipped to deal with collisions.
4 laptops using wifi, 4 TVs using wifi, 6 mobile phones using wifi, 4 headphones using bluetooth, 100 home automation products using zigbee.
Wifi network will typically carry a lot of data and zigbee network will be carrying less amount of data but at frequent intervals.
Is such a network sustainable in the long run from RF interference perspective?
Yes. Experience tells us this is not a problem, especially since the wifi devices will tend to use the 5 GHz bands to get more bandwidth, anyway, leaving the farther-reaching, potentially lower-rate 2.4 GHz band free for the other systems. Anyway, zigbee very much was designed with such scenarios in mind and should be robust enough. As said, law requires your Wifi devices to leave enough space!
A wifi router supporting 5-10 devices is one thing but will it translate successfully to 100-200 devices?
completely different problem. You won't be connecting 200 devices to a single AP; you will have multiple, cooperating (if possible, even coordinated) APs distributed locally so that each wifi device only needs the minimum possible transmit power – and hence, interference becomes much more limited.
There's a nice corollary to "the more central nodes I can distribute throughout space, the less power my individual mobile nodes will need to communicate with the infrastructure": The more cell towers you put up, the lower the average user's average exposure to RF energy becomes. That's why people protesting against cellular base stations in their neighborhood because they're afraid of being exposed to microwave radiation are counter-productive to their own goal.
The magic word you're looking for here is "network densification": If I serve a large area with just one central node, then, well, the spectrum of that area has to be split among all users. If you make your cells smaller, by using less power, more central nodes, and higher frequencies, you can use the same spectrum in different places at the same time – bingo, higher total throughput.
