Route A[sym, :] through A[sym] for symbolic indices

`sol[arr_sym, :]` should be equivalent to `sol[arr_sym]` — both ask for
the full timeseries of the symbolic indices. Currently the `:`-suffixed
form errors with a broadcast shape mismatch inside SII's `GetStateIndex`
when the underlying index resolves to a `Vector{Int}` (i.e. when the
symbol is array-valued like `x(t)[1:3]`):

    DimensionMismatch: arrays could not be broadcast to a common size:
    a has axes Base.OneTo(N) and b has axes Base.OneTo(K)

(SII's `(gsi::GetStateIndex)(::Timeseries, prob, i)` broadcasts `gsi.idx`
against `state_values(prob, i)` without wrapping it in a 1-tuple, so a
length-N timeseries broadcast against a length-K idx vector errors. A
companion fix is being submitted upstream.)

Short-circuiting `A[sym, :]` to `A[sym]` here both restores the intended
semantics and avoids the broken SII codepath.

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>

Author: ChrisRackauckas

src/vector_of_array.jl

@@ -797,6 +797,15 @@ Base.@propagate_inbounds function Base.getindex(A::AbstractVectorOfArray, _arg,
     end
     symtype = symbolic_type(_arg)
     elsymtype = symbolic_type(eltype(_arg))
+    # For symbolic indices, `A[sym, :]` is semantically equivalent to `A[sym]`
+    # (the no-args symbolic getindex already returns the full timeseries).
+    # Routing through the no-args path here also avoids a broadcast shape bug
+    # in SymbolicIndexingInterface's `GetStateIndex` when the underlying index
+    # is a `Vector{Int}` (array-symbolic) combined with a `Colon` time index.
+    if (symtype != NotSymbolic() || elsymtype != NotSymbolic()) &&
+            length(args) == 1 && args[1] === Colon()
+        return A[_arg]
+    end
 
     return if symtype == NotSymbolic() && elsymtype == NotSymbolic()
         if _arg isa Union{Tuple, AbstractArray} &&