Fiber splicers are marvels of technology. They align the fiber cores with sub-micrometer accuracy and produce just the right amount of heat and pressure to melt the ends together. They are also usually very rugged, fully automated, and surprisingly cheap (a few thousand euros). It is remarkable what is possible when the entire internet relies on a technology.

That is probably the very best case scenario, but possible yes. Typically you'd accept anything less than 0.1dB.

I did my PhD on fibre lasers, 0.1 DB would have been considered a ver bad splice and I would have recut and respliced (if you have 1-10W in your cavity that 0.1 dB loss would risk burning and the fuse propagating through your cavity destroying everything in its path (as a side not look up Videos of fibre fuse, looks fascinating). In my experience 0.01-0.02 is much more typical than 0.1 dB loss.

I’m speaking mainly within the context of telecom field splicing - the numbers I mentioned are typical for that application in my experience. You’re only sending on the order of 5 mW down a fiber, so none of those high-power concerns apply. Obviously, different networks have different thresholds: if you’re building a greenfield, low-latency long-haul route, you want to minimize loss and it’s reasonable to spend the extra time and use higher-end equipment. For FTTH, with something like a 30 dB overall budget, nobody really cares whether a splice is 0.03 dB or 0.1 dB.

You might accept a bad splice but you'd almost have to fuck it up on purpose with a decent automatic splicer.