Should you kiss it better?

I don't think I see a problem with this situation.  "Do you want me to kiss your injury?" is simply asking about their preferences, and makes no assertion about the healing powers of the kiss.  "Do you want me to kiss it better?" does make a bit of an assertion, but at least at this point, it's clear that it provides comfort to your son—it makes him feel better—so I think the phrasing is ok.

Also, there is research suggesting that saliva on wounds is in fact helpful:

Our results show that human saliva can stimulate oral and skin wound closure and an inflammatory response. Saliva is therefore a potential novel therapeutic for treating open skin wounds.

Licking wounds is useful elsewhere in the animal kingdom; see Wikipedia.  My guess for humans would be "it probably helps a bit—though, if Neosporin is available, I'd prefer the latter".  If it does help, then I think we'd expect baby animals who aren't able to lick the wounded area (or who just don't lick it for some reason) to be adapted to, upon getting injured, cry loudly until a parent comes by and licks their wounds / otherwise attends to them; and humans probably inherited the adaptation.

I'd say it's less likely (though possible, with the "inflammatory response"?) that it helps with bruises or other injuries that don't break the skin.  But it's also plausible that babies can't tell the difference between a wound that bleeds and one that doesn't, or that evolution didn't find it worthwhile programming an exception for the latter.

So I suspect that your son's crying followed by his "it's magically better now" response may be an evolved behavior.  (Even if saliva did nothing, the general "cry for help when you get hurt until someone tends to your wounds" reaction seems reasonable.)  In which case kissing his injury is speaking directly to that evolved part of him.

The dishonesty isn't in "It'll be okay after I kiss it", it's in the idea that it was ever not okay in the first place. 

Think about what purpose the idea that kissing it makes it better serves to you, which leads you to use it. The kid won't accept "Aw, you're fine" and therefore you can't help him feel better with that take. However, if you pretend that he really is not okay, then he can believe that, and maybe he'll also believe you when you offer a solution. It allows you to side step the part where you have to convince the kid that he's wrong, and allows you to lead him to a correct conclusion that he's okay. Which is nice, since having your toddler hurt and in distress can be uncomfortable, and we'd often like to get out of that discomfort ourselves.

But that work is where all the cool stuff happens, and that's where you get to teach them the skill of reorienting to unpleasant sensations effectively. The alternative starts with orienting towards our own discomforts skillfully, and then modeling that for them as applied to their problem. 

So, like, what happens when your kid scrapes his knee and he cries about it, and he doesn't stop crying for five minutes. When you could have stopped it at two with a kiss. Is that okay, and something you feel comfortable playing with (assuming you see a reason to play with it), or is it something uncomfortable which you'd rather stop?

To the extent that it's the latter then you have you own little puzzle to sort out, and to the extent that it's the former then you have a new game to play with your toddler. When your own emotional take on the booboo and ensuing distress is "Ooh! An opportunity to play! How bad is this one!?", then it tends to come across and the kid can learn that little booboos and a little distress aren't the end of the world and can actually be a fun learning challenge in an interesting sort of way. You get to engage with the experience they're having without trying to minimize it, or to pretend to agree with it, and that gives them a lot more room to figure out if they're actually okay and what lessons (if any) they want to take from it. And you often get really cool experiences, like watching your kid reframe the problem from "I'm hurt and not okay" to "It was scary and I cried, but I'm okay!" and "Can we do it again!?".

In principle, this can result in more distress because there's more willingness to entertain distress, but in practice I haven't found it to be the case. There aren't really any times where I could offer a kiss to make it feel better, because in the cases where it'd work it never really becomes a problem in the first place. Sometimes my toddler will insist that she needs a bandaid to make it better, and I'll give her one even though she doesn't physically need one, but that's very much led by her and the nudges she gets from me are actually away from the idea that the bandaids are necessary.

I agree with others that it's not really lying.

But also, it doesn't matter whether you're lying to them. It's not like lying is some sin against God. The relevant thing here is, What impact are you concerned about?

It's not like you're deceiving your child about cause-and-effect. They hurt, you kiss to make the pain go away, and the pain often goes away. That's quite real.

Where exactly is the problem?

In my mind, if this is "a principled exception" to some rule, then your rule needs updating. It doesn't reflect reality. What you're doing with your kid here is obviously fine.

I think it's something of a frame mistake to talk about lying when the other person is not physically capable of understanding the truth, and often is still learning the concepts he will use to assess truth in the future. If a child is incapable of understanding the physics of river currents, telling them "a monster will suck you under the water" is less a lie and more a translation. If a child says "I'm hot" when it's freezing and you suggest "I'm cold", are you gaslighting him or expanding his vocabulary?

"kiss it better" could be an aggressive lie, or it could be the most understandable way to convey to him that you acknowledge and care about his pain, without making it bigger than it is. It also empirically seems like it doesn't override his own sense of "no, this actually hurts", which is the most obvious potential cost.