Loong

We have run their tests before 0bcb405ca4efd3844d172a5516765422d0bbc84d, but it's removed due to #184. It seems works well now, but it increase the compile time significantly (The full tests of `LanguageServer.jl`...

IMO, We can update `Manifest.toml` by github actions like: 1. Clone `julia-vscode` and sync submodules. 2. Get the commit hash of `LanguageServer.jl` in `julia-vscode`. 3. Update `Manifest.toml` with the commit...

I think that `vscode-julia` switch environment by [send notification to LS]( https://github.com/julia-vscode/julia-vscode/blob/59e4c3ee65ee4e5c66e7a99e7cc297f72437c856/src/jlpkgenv.ts#L108). I try it in nvim by ``` :call CocNotify('julia', 'julia/activateenvironment', '/Users/loong/.julia/environments/v1.7') ``` However, the LS restart but don't...

I think there are some issue about diff coverage. Lines warned by codecov have been covered.

I tried to make `getindex(::SubCartesianIndices)` returns `NDIndex`. But it only for `ArrayIndex.getindex` now, because making `Base.getindex(::SubCartesianIndices)` returns `NDIndex` will break tests for `CartisianIndices`.

@Tokazama Can we extend the `static_unsafe_getindex` as definition of `ArrayInterface.getindex` for ranges? It might be useful to preserve `StaticInt`.

The `Base.getindex` will always returns a `Int`, but it will preserve it as a `StaticInt` when possible. ```julia julia> (static(1):static(2))[static(1)] 1 julia> ArrayInterface.static_unsafe_getindex(static(1):static(2), static(1)) static(1) ```

> Maybe we shouldn't worry about forcing this to work with `CartesianIndices` and we just want our own type that functions as a more efficient replacement? So we should't make...

> It seems like we might actually just want to replace `CartesianIndices` with this new type entirely so we don't have to hack our way around so many issues. If...

@Tokazama, thanks for your reply, we can continue this PR when you are free.