Fix downstream test breakages exposed by OrdinaryDiffEq 7 bump

The compat bump in the previous commit gets `activate_downstream_env`
past `Pkg.develop`, but the downstream test sources still don't run
because the OrdinaryDiffEq 7 resolution was previously masking three
issues:

- `Rosenbrock23(autodiff = false)` is no longer supported; OrdinaryDiffEq 7
  rejects the `Bool` form and asks for an `ADType`. Use `AutoFiniteDiff()`
  (already reachable through `using OrdinaryDiffEq`).
- `Rodas5` is no longer re-exported from `OrdinaryDiffEq`; pull it in
  via `using OrdinaryDiffEqRosenbrock` and add the package as a dep.
- `AP[...]` shorthand requires `RecursiveArrayToolsShorthandConstructors`
  in scope; runtests.jl `Pkg.develop`s the package but never `using`-s it,
  so the downstream test files silently never had it.
- `SciMLBase.successful_retcode(...)` requires `SciMLBase` as a bare
  binding, which `using OrdinaryDiffEq` doesn't provide. Switch to the
  unqualified `successful_retcode` (also already exported through
  OrdinaryDiffEq), avoiding the need to add SciMLBase as a direct dep.

Verified locally that the ODE Solve / Events / Measurements and Units /
TrackerExt safetestsets all run to completion after the fix.

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

Author: ChrisRackauckas

test/downstream/Project.toml

@@ -4,7 +4,9 @@ ModelingToolkit = "961ee093-0014-501f-94e3-6117800e7a78"
 MonteCarloMeasurements = "0987c9cc-fe09-11e8-30f0-b96dd679fdca"
 NLsolve = "2774e3e8-f4cf-5e23-947b-6d7e65073b56"
 OrdinaryDiffEq = "1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"
+OrdinaryDiffEqRosenbrock = "43230ef6-c299-4910-a778-202eb28ce4ce"
 RecursiveArrayTools = "731186ca-8d62-57ce-b412-fbd966d074cd"
+RecursiveArrayToolsShorthandConstructors = "39fb7555-b4ad-4efd-8abe-30331df017d3"
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 SymbolicIndexingInterface = "2efcf032-c050-4f8e-a9bb-153293bab1f5"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
@@ -18,6 +20,7 @@ ModelingToolkit = "8.33, 9, 10, 11"
 MonteCarloMeasurements = "1.1"
 NLsolve = "4"
 OrdinaryDiffEq = "6.31, 7"
+OrdinaryDiffEqRosenbrock = "1, 2"
 StaticArrays = "1"
 SymbolicIndexingInterface = "0.3"
 Tables = "1"

test/downstream/downstream_events.jl

@@ -1,4 +1,4 @@
-using OrdinaryDiffEq, StaticArrays, RecursiveArrayTools
+using OrdinaryDiffEq, StaticArrays, RecursiveArrayTools, RecursiveArrayToolsShorthandConstructors
 u0 = AP[SVector{1}(50.0), SVector{1}(0.0)]
 tspan = (0.0, 15.0)
 

test/downstream/odesolve.jl

@@ -1,4 +1,5 @@
-using OrdinaryDiffEq, NLsolve, RecursiveArrayTools, Test, ArrayInterface, StaticArrays
+using OrdinaryDiffEq, OrdinaryDiffEqRosenbrock, NLsolve, RecursiveArrayTools,
+    RecursiveArrayToolsShorthandConstructors, Test, ArrayInterface, StaticArrays
 function lorenz(du, u, p, t)
     du[1] = 10.0 * (u[2] - u[1])
     du[2] = u[1] * (28.0 - u[3]) - u[2]
@@ -9,7 +10,7 @@ u0 = AP[[1.0, 0.0], [0.0]]
 tspan = (0.0, 100.0)
 prob = ODEProblem(lorenz, u0, tspan)
 sol = solve(prob, Tsit5())
-sol = solve(prob, AutoTsit5(Rosenbrock23(autodiff = false)))
+sol = solve(prob, AutoTsit5(Rosenbrock23(autodiff = AutoFiniteDiff())))
 sol = solve(prob, AutoTsit5(Rosenbrock23()))
 
 @test all(Array(sol) .== sol)
@@ -72,4 +73,4 @@ end
 u = fill(SVector{2}(ones(2)), 2, 3)
 ode = ODEProblem(rhs!, VectorOfArray(u), (0.0, 1.0))
 sol = solve(ode, Tsit5())
-@test SciMLBase.successful_retcode(sol)
+@test successful_retcode(sol)