It is not necessary really.

There is unlikely any non-state actors[1] that has the ability to transmit signals to L2 . Just receiving signals even now (only 2 out of 30 days to l2) the OP used a 6 meter dish. Most of interplanetary mission signals are handled by the DSN.

Any sort of encryption will add both b/w requirements and compute requirements . The CPU/network budgets on such missions are very very limited. Every bit and cycle counts.

Finally standard encryption libraries, algorithms et al, are not likely suitable . I am no expert, but I have not read of any modern algorithms with very low network overhead + compute requirements designed for these kind of use cases, that is also secure from brute force or other attacks.

Mission risk is also a factor, even handshake failures can jeopardize the mission. It is one thing a website did not load because of TLS negotiation failures and $10 B mission overshot its orbit because handshake failures on the encryption layer.

[1] Threats from state actors for science missions is different category of concern, harder to quantify and with not much history of actual attacks. Collateral risks like from the ASAT Russian test to ISS, or in dual use missions would perhaps not apply here .Usually science teams collaborate well even if there is lot of tension in political sphere.

"It is not necessary really."

Authentication of commands to satellites is very, very necessary

Encryption !=authentication. OP was talking about encryption.

You could do authentication over plain text. For popular example http basic auth.

It is not recommended for regular use cases, but is not out of realm of possibility in satelite given the constraints.

OP was talking about digital signatures and a hobbyist taking command of the spacecraft. In other words authentication.

You said that was unnecessary and too taxing on constrained hardware. That is incorrect. Authentication is both necessary and not excessively taxing, when both considering the risk to the spacecraft's operation or even not considering it, since, as you said yourself, authentication schemes can be reasonably lightweight.

"Not out of the realm of possibility in satellites" What are you talking about? Of course it isn't! Authentication to satellites is recommended and implemented all the time, for obvious reasons. You think they risk a multi million dollar investment to save some clock cycles? How many commandeered satellites do you read about daily? Do you have sources to back up any of this?

I'd imagine the specs are rated and hardened for radiation first, as seen on all previous NASA satellite and probe missions before getting into the weeds of overhead and encryption.

"Torpedo in the water!" would probably be sufficient.

The US government being able to discreetly torpedo everyone everywhere within what would have to be a few minutes at most, that would be pretty pretty scary. Imagine the number of drones they'd need to have deployed and armed at all times, and the potential for abuse.

I think you might just be trying to make this threat a much bigger issue than it really is. How many people do you really think will be trying to attack the JWST?