fix invalidations in REPL LineEdit.jl from Static.jl

[ranocha]

This is based on the following code:

julia> using Pkg; Pkg.activate(temp=true); Pkg.add("Static")

julia> using SnoopCompileCore; invalidations = @snoopr(using Static); using SnoopCompile

julia> trees = invalidation_trees(invalidations)
inserting !(::False) in Static at ~/.julia/packages/Static/sVI3g/src/Static.jl:427 invalidated:
...
                 75: signature Tuple{typeof(!), Any} triggered MethodInstance for REPL.LineEdit.var"#add_nested_key!#24"(::Any, ::typeof(REPL.LineEdit.add_nested_key!), ::Dict, ::Union{Char, String}, ::Any) (60 children)
...

I suggest the labels latency and backport-1.8.

[dkarrasch]

I'm not at all familiar with the invalidation business, but judging from the recurrent patterns: the negation acts on the two expressions whose final value is passed to !, so are you sure you don't need

Suggested change

	elseif !((c in keys(keymap))::Bool && isa(keymap[c], Dict))
	elseif !((c in keys(keymap) && isa(keymap[c], Dict))::Bool)

[ranocha]

I checked with Cthulu and the problem was that c in keys(keymap) was inferred as Any.

[JeffBezanson]

Looks like there is an opportunity to improve inference here:

julia> code_typed(Base.KeySet, (Dict,))
1-element Vector{Any}:
 CodeInfo(
1 ─ %1 = $(Expr(:static_parameter, 1))::Any
│   %2 = $(Expr(:static_parameter, 2))::Type{T} where T<:Dict
│   %3 = Core.apply_type(Base.KeySet, %1, %2)::Type{Base.KeySet{_A, _B}} where {_A, _B}
│   %4 = (%3)(dict)::Base.KeySet
└──      return %4
) => Base.KeySet

If that returned KeySet{<:Any, <:Dict} we would know the exact in method and it would get inferred.