Requesting an image of a paranoid person's server isn't necessarily that great. When I worked for a run-of-the-mill cybersecurity firm, our simulator products were protected with full disk encryption using run-of-the-mill open-source software + light patches and keys bound to specific hardware, software, and configuration states via the TPM. This is for fully automated boot up. If you can accept the risk of needing to be physically close to a machine, you can generate random bytes and store those into your TPM and require both the hardware/software/configuration to be correct as well as knowing your key. This would incidentally also prevent you from being able to give law enforcement the key to an image of your computer (this is actually impossible, you don't know the key).
If you're doing this under a warrant, you could just request that the server's operator unlock the machine. Whether you comply is a legal situation that varies from jurisdiction to jurisdiction (in the US, it seems that you might be held indefinitely in jail if you refuse to divulge your key). The thing is, you should be able to make an extremely strong case (possibly with the EFF's help) that any warrant is false. Anonymous traffic itself should not be enough to compel you to divulge your secrets without other evidence pointing to your machines (standard IANAL, but this seems consistent from everything I've read).
If you're doing this under a warrant, you could just request that the server's operator unlock the machine. Whether you comply is a legal situation that varies from jurisdiction to jurisdiction (in the US, it seems that you might be held indefinitely in jail if you refuse to divulge your key). The thing is, you should be able to make an extremely strong case (possibly with the EFF's help) that any warrant is false. Anonymous traffic itself should not be enough to compel you to divulge your secrets without other evidence pointing to your machines (standard IANAL, but this seems consistent from everything I've read).