Remove ForwardDiff hack from tests

This was previously needed because of
<SciML/OrdinaryDiffEq.jl#2900>, which was fixed by
<SciML/DiffEqBase.jl#1219>.

Author: giordano

test/Project.toml

@@ -3,7 +3,6 @@ Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
 ChunkSplitters = "ae650224-84b6-46f8-82ea-d812ca08434e"
 DelimitedFiles = "8bb1440f-4735-579b-a4ab-409b98df4dab"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
-ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 GaussianRandomFields = "e4b2fa32-6e09-5554-b718-106ed5adafe9"
 HDF5 = "f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
@@ -21,7 +20,6 @@ Aqua = "0.8"
 ChunkSplitters = "3.1"
 DelimitedFiles = "<0.0.1, 1"
 Distributions = "0.22, 0.23, 0.24, 0.25"
-ForwardDiff = "1.2"
 GaussianRandomFields = "2.2.1"
 HDF5 = "0.14, 0.15, 0.16, 0.17"
 LinearAlgebra = "<0.0.1, 1"

test/models/lorenz63.jl

@@ -5,8 +5,6 @@ using Distributions
 using HDF5
 using Random
 using PDMats
-# `ForwardDiff` is only needed to trigger loading a pkgextension.
-import ForwardDiff as _
 using OrdinaryDiffEqTsit5: OrdinaryDiffEqTsit5, Tsit5, ODEProblem, reinit!, step!
 using ParticleDA