Could the university you work for provide you with a second machine whilst the first machine runs experiments?
Also there will always be distractions; daydreaming if nothing else. At a previous employer my main work machine was not connected to the Internet (a nightmare for other reasons, like if I was on the Internet-enabled machine, may as well check email...). But even then, when waiting for a test that would take several minutes, I'd find myself reading a print newspaper.
These forced delays kill productivity for me like nothing else. Losing the task at hand every couple of minutes to retest or really slows down output.
Recently, though, I've stop listening to most of the "productivity advice" and warnings about multitasking, splitting your time into chunks, and the myriad of other approaches.
The best bit of advice I have heard is just to get on with it, work hard, and stop procrastinating about it. Too much meta is not good.
Sure. They do, in fact. And I use clusters for large experimentation.
However, a great rule of thumb as a researcher is to try to limit yourself to using a single laptop for your work and experiments. One advantage: your development machine and experimental machine are then identical, and they both travel with you. Academics travel a lot.