Would it be possible to use weapon projectiles instead of hitscan as a way of reducing the perception of network latency in UE4 / eve:fps?

Sort of
Client: Press button -> Fire Animation -> Spawn Projectile -> Hit!
Server: ............................Server receives and compensates...Hit!

Instead of
Client: Press button, hit! -> Render hit
Server: ............................Server receives and compensates...Hit!

(The idea comes from Tekken, stuff here)

More precisely I think you're referring to bullet travel time rather than visible projectiles.

One issue with visible projectiles is that the server can't perform lag compensation on projectiles that are spawned on the client side as the projectile would visibly deviate from the intended trajectory. So they have to be spawned server-side, which puts the lag between L1 and the spawning of the projectile (refer to the Forgegun as an example of this). Alternatively you can keep the visible projectiles client side but the impact calculations server side, in which case you get the same difference between visible impact and game mechanics you get with hitscan weapons (refer to the Scrambler Rifle as an example of this).

What you may be referring to is something like the half-life server model. I don't actually know what it's called in technical terms, but the advantage is that regardless of the amount of latency the client and server will always agree on what the client saw when it took an action and what effect it should have. It accomplishes this by having the server keep a store of what it told each client about where everyone was at the time the client took an action and performing the hit detection on that dataset. The downside is a fairly big memory- and cpu-footprint (plus mental effort on the side of the programmers!). Cheating through intentionally bad latency is also a possibility.

maybe in project legion but dust will have no further updates.

More precisely I think you're referring to bullet travel time rather than visible projectiles.

One issue with visible projectiles is that the server can't perform lag compensation on projectiles that are spawned on the client side as the projectile would visibly deviate from the intended trajectory. So they have to be spawned server-side, which puts the lag between L1 and the spawning of the projectile (refer to the Forgegun as an example of this). Alternatively you can keep the visible projectiles client side but the impact calculations server side, in which case you get the same difference between visible impact and game mechanics you get with hitscan weapons (refer to the Scrambler Rifle as an example of this).

What you may be referring to is something like the half-life server model. I don't actually know what it's called in technical terms, but the advantage is that regardless of the amount of latency the client and server will always agree on what the client saw when it took an action and what effect it should have. It accomplishes this by having the server keep a store of what it told each client about where everyone was at the time the client took an action and performing the hit detection on that dataset. The downside is a fairly big memory- and cpu-footprint (plus mental effort on the side of the programmers!). Cheating through intentionally bad latency is also a possibility.

maybe in project legion but dust will have no further updates.