I had one but sold it. I bought it used around the same time I bought a UR10e. Ended up keeping the UR because of how easy it is to use. The guy using the Epson seems to be doing well with it.
Durability of the Epson appeared to be quite good. It was very beefy and heavy for a 6kg robot. Not sure about its IP rating but I think it can take a splash.
That said, if you're looking at a turnkey solution, I'd rethink that approach and instead DIY the integration, either with the Epson or a cobot. That critical first robot is going to pave the way for the rest of your future automation ambitions. IME, a bad turnkey can really leave a sour taste in your mouth.
Once you figure out general robot movements, you're inevitably going to shift focus towards problems in your automation system that have nothing to do with the robot itself, but will require robot reprogramming. For example, chip contamination in the workholding that might require the robot to remove the part, the door to close, have the machine run a self cleaning cycle (coolant + airblast + chip fan), and then open the door again. The idea that a pitiful air nozzle connected to the robot can solve all chip problems is wishful thinking at best. Most robot integrators aren't machinists and don't get it.
If you program your robot yourself from the start, it'll be much easier to tweak that program later. If someone else programs your robot from the start, it's going to be an uphill battle figuring out what the heck he did and trying to to fill the gaps in your knowledge.