If you want an off-the-shelf drone, besides the fact that we tend to not provide recommendations for specific hardware, I'm not sure there even is one like that in the hobbyist market. Maybe you could find an industrial-grade solution, but it's unlikely that it is to be found on a hobbyist-focused site like this one.
You could, however, build something of the sort without too much effort.
The first three points can more or less be satisfied with a generic hobbyist-grade flight controller flashed with an (also generic) autopilot-enabled firmware.
Of the ones freely available for hobbyist use a considerable number have the capability of operating autonomously without a control link. AFAIK, all of the most popular ones offer it in some form.
Point two is a bit harder than the rest; the flight control firmware will need to have the capability for extending its behaviour.
This can take either the form of onboard scripting, where the custom code runs on the FC directly, or external control by a second computer (e.g. Raspberry Pi) mounted on the same aircraft and communicating with the FC. There's also an option to compile the intended behavior into the software itself, but that takes quite a bit more effort.
See the answers to this question for a more detailed overview of such capabilities in different autopilot packages.
Point three is satisfied more or less automatically: about any flight controller has the standard binary interfaces (UART, I²C, SPI) for communicating with various sensors and actuators, and a bunch of GPIO pins for other uses, and about any flight control firmware can use all of them. In addition, if you decide to go the two-brain route (i.e. an FC running the basic autopilot interfaced with an SBC running linux and any custom logic you might need), the SBC will have even more interfaces, including USB, and quite a bit more processing power to tap into than the FC microcontroller has.
The next two points are more about the airframe and hardware design; You'd be best served browsing for answers/asking the questions separately (e.g. How do I make a drone sturdy / how do I achieve long flight times); both of these have been discussed extensively all over the RC/FPV hobbyist communities. Or you could just search for a sturdy long-flying build separately, buy that and add your autopilot package on top.
Finally I must address the EMP shielding query.
Unless you plan to overfly military installations or similarly-equipped forces, it's unlikely you'll have to worry about full-blown EMP. In fact, it's probably true even if you do plan to do that (which I don't recommend).
If you're worried your drone might be interfered with, I'd suggest protecting yourself from two more specific kinds of electronic countermeasures: comms jamming and GPS spoofing (which are both illegal in most places, but that doesn't stop people from using them in an attempt to defend from drones).
The first should be more or less solved by the comms-independent routines you're suggesting, but having redundant comm channels is a good idea anyway; You're much more likely to have to deal with natural communication dropouts than manmade ones.
GPS spoofing is more dangerous, as it makes your GPS receiver report a false location. Most likely the fake location will be at an airport (which is a no-fly zone and will cause commercial drones to land immediately). Self-built drones usually don't have a database of such locations anyway, so the main danger here is that it'll cause it to lose any sense of location and, unless you've got a backup navigation system on your drone, make it unable to return home until it leaves the spoofed area.
This is a hard problem that, as far as I know, is not solved in any of the hobbyist autopilots in any way (most likely because their developers don't live in areas where such spoofing is common), so you're on your own here. If your area is not covered in spoofers like mine is, though, a simple detection and return-to-home routine might be enough to avoid losing the aircraft to this.