How to ensure the fault tolerance of the DGS-3120 switch?

Question

Computer networks

Tell me, please, what can be done with such a D-Link DGS-3120-24TC/RI switch https://www.dlink.ru/ru/products/1/1782.html
It is in the "core" of a small network, distribution switches from other buildings and floors are connected. RSTP is not used. This is what it looks like.


I wish this switch wasn't a network failure point. There are 2 more of the same switches in stock (not used now). Do not just offer to change it to another switch, because this is already understandable, but it costs money.

DGS-3120 does not support stack in RI version, only in EI. And RI is needed, because it routes directly connected IP networks of different vlans.
Also it does not support M-LAG.

My suggestion is to apply ERPS (faster responsive than RSTP) by connecting three 3120s in a ring. How then to be with distribution switches? Just split them across those three switches? The fall of a third of the network is still better than the whole.
Or add one 3120 to the existing one, add redundant links from distribution switches and configure RSTP? More precisely MSTP, since vlans are used here.

What is the best way to achieve fault tolerance if this model does not give anything?