improve logging in semantic tokens rule

[soulomoon]

Switch use_ to useE to capture meaningful error.

[michaelpj]

Do we need to do this manually everywhere? What does this do, concretely?

[michaelpj]

(I am pretty confused about how all this works tbh!)

[soulomoon]

Do we need to do this manually everywhere? What does this do, concretely?

If we can print out the throwed error in use_ in the general handling of rules, then we do not have to do it everywhere,
It must be captured some where and hidden. But tbh, I am not sure where it is.

[soulomoon]

BadDependency is swallow here Is around here.

pure (Nothing, ([ideErrorText file $ T.pack $ show e | not $ isBadDependency e],Nothing))