The TRESOR kernel patch is for x86 processors, as it uses the x86 debug registers to store private key information (at the cost of disabling hardware debugging). The Raspberry Pi is based on ARM architecture, which is completely different.
I'm not sure you actually gain anything from running TRESOR on a Raspberry Pi in any case - even if it was supported - because the memory on a Pi is soldered to the board. It's not user-replaceable.
Frankly, if you don't trust locks, there isn't much point in installing anything of interest on the site itself: You'd be better off setting up a router, and having the (internet-enabled - maybe streamed via your Pi?) cameras stream content to a server hosted on the internet somewhere. Maybe an Amazon instance, maybe something else in the cloud: How you want to do that is up to you. At the end of the day, TRESOR or not, if someone got in and simply stole your Pi and hard disk, you'd never know who did it, because they'd be holding the only digital evidence. But if that same someone came in and unplugged your router - and got caught on camera in the process, you'd at least have some recorded evidence that you could use.
(Assuming they didn't trip the power breaker for your apartment, first...)
You need to decide what you want, here: TRESOR solves all the wrong problems. If you still want to pursue defence against cold-boot attacks, despite what has been said here, you could take a look at ARMORED, which you could simply view as a version of TRESOR for ARM architectures.