Aligning docstrings and comments

[domluna]

@ZacLN

Currently the formatter doesn't properly align docstrings and comments.

begin

"""
doc for `foo`
"""
function foo()
    return 10
end

end

formats to:

begin

    """
doc for `foo`
"""
function foo()
    return 10
end

end

I have a POC which treats each line in the doc as a LITERAL instead of the entire doc as one LITERAL and this works, also needs come changes to get_lines. We could do something similar for comments but it's trickier since at the moment CSTParser ignores comments.

We could add something like

abstract type Comment <: Head end
abstract type MultiLineComment <: Head end

to spec.jl and then store the comments as LITERALS, MultiLineComment would be similar to how docstrings work.

I have some WIP on these things but would like to hear your thoughts before I proceed further.

[ZacLN]

I don't think modifications to CSTParser are the correct route for this, having a quick glance I think we should be able to hand this through modifying get_lines. I think handling comments explicitly in CSTParser would add too many complications to the token look ahead mechanism