this is rather a simple and fundamental question but nevertheless i couldn't find any solution for it and now I'm worried if it can cause real attacks to slow blockchains.
imagine in a form of committee some users are chosen to propose blocks. for some reason this committee stalls to send the blocks to the rest of the network just until a few milliseconds before the time window is ended. for the few nodes near them, the blocks are sent in the right time window but for the rest of the network they received the block after the window is ended. now we have a situation where some nodes say this block is valid and propagated in the tight time span but some users saying otherwise because they received a block in an invalid time. therefore to overcome this issue one way that comes to mind is to stall the rest of the network and accept the late delivered block in which case the network can be easily slowed down constantly. is there any point I'm missing here or is this a real concern? how is it that ethereum claims this would not happen in its network once a peer asks a trusted source (weakly subjective)?
also this problem in tron is refereed to as temporary fork. so in a simpler manner how does the network recover from temporary forks due to partitioning and keeping the finality of the transactions happened in both forks as they propose they do?