Type changes when using faces to index points

When defining a face vector F and a point vector V one could access the points for face 1 like this:

V[F[1]]

I've noticed that the type returned by the above is first of all not the same type as V (which would be desirable for my applications, but perhaps you have reasons to change the type) but also that the type depends on the face integer type.

For instance here using Int32:

T = Int32
F = TriangleFace{T}[ [1,2,3] ]
V = Point{3,Float64}[ [0.0,0.0,0.0], [1.0,0.0,0.0], [1.0,1.0,0.0]]

vf = V[F[1]]
println(typeof(vf))

The type of vf is: StaticArraysCore.SVector{3, Point{3, Float64}}
Whereas when using Int64,

T = Int64
F = TriangleFace{T}[ [1,2,3] ]
V = Point{3,Float64}[ [0.0,0.0,0.0], [1.0,0.0,0.0], [1.0,1.0,0.0]]

vf = V[F[1]]
println(typeof(vf))

The type of vf is: StaticArraysCore.SizedVector{3, Point{3, Float64}, Vector{Point{3, Float64}}}.

The returned types should probably be the same right?

Side note, V[F[1][1:end]] returns a Vector of Points so is different from V[F[1]] again. Would it not be best to always just return a vector of Points?

Reproduce the reported behavior with the Int32 and Int64 examples using V[F[1]], then compare it with V[F[1][1:end]]. Trace the face-indexing and point-vector indexing entry points to identify where the return type diverges. Done means the supported indexing forms have a consistent, documented return type and regression coverage for both integer types.