meta: suggestion: require govet for linting

[ValarDragon]

Do we want to use go vet in the linter? Here are some sample things it wants to fix:

		accAuth.Coins = sdk.Coins{
			{genTx.Name + "Token", sdk.NewInt(1000)},
			{"steak", freeFermionsAcc},
		}

to

		accAuth.Coins = sdk.Coins{
			{Denom: genTx.Name + "Token", Amount: sdk.NewInt(1000)},
			{Denom: "steak", Amount: freeFermionsAcc},
		}

(The objection here was only because the struct definition was nested)

And it noticed that some json tags were broken, we had:
client/tx/sign.go:14::error: struct field tag json="name"``
it should have been json:"name".

I think it may be useful, but I don't feel strongly about it. (The main benefit being it catching the second bug) We currently spend alot of reviews making nitpicks on stuff that a linter in principle could catch,

Here's the golang page describing govet: https://golang.org/cmd/vet/

Optionally, we can turn off its composite literal check, since I think thats the least impactful of the things it provides. (I personally like it, but I don't want that being annoying to detract from what else it can help with)

[ValarDragon]

Closing in favor of 1417