Minor performance regressions on v0.7

[ChrisRackauckas]

Mostly due to JuliaLang/julia#28126 . This is the current state.

Non-stiff 100x 100x linear:

> wp[1]
Name: DP5
Times: [0.0011027, 0.00128265, 0.00171914, 0.0021727, 0.00310171, 0.00463389, 0.0070847, 0.0111793, 0.016864, 0.0262313, 0.0415285]
Errors: [1811.17, 1068.78, 141.098, 0.599027, 0.944153, 0.138538, 0.0158417, 0.00171682, 0.000178345, 1.81836e-5, 1.84698e-6]


> wp[2]
Name: dopri5
Times: [0.00246687, 0.00331948, 0.00403105, 0.00582659, 0.00820785, 0.0131199, 0.0194768, 0.0307647, 0.0492491, 0.0750954, 0.116532]
Errors: [2967.81, 1130.19, 120.829, 0.348979, 0.93898, 0.139212, 0.0158875, 0.00171261, 0.000178078, 1.82033e-5, 1.84583e-6]


> wp[3]
Name: ARKODE
Times: [0.0194515, 0.0198796, 0.0213923, 0.0230518, 0.0270304, 0.0294924, 0.0374738, 0.0617449, 0.0813931, 0.120038, 0.14512]
Errors: [772.501, 62.8917, 1.1652, 1.41221, 0.235034, 0.087287, 0.00840397, 0.000260545, 2.41061e-5, 2.17281e-6, 7.74874e-7]


> wp[4]
Name: Tsit5
Times: [0.001435, 0.00165658, 0.00223777, 0.00279761, 0.00390899, 0.00538023, 0.00793093, 0.0120528, 0.0181457, 0.0274027, 0.0430627]
Errors: [410.046, 138.615, 27.8208, 3.03454, 0.248046, 0.0119785, 5.55002e-5, 7.47631e-5, 1.26288e-5, 1.59618e-6, 1.80577e-7]

Pretty much no regression.

Stiff: HIRES

#0.7

Name: Rosenbrock23
Times: [0.000410643, 0.000599607, 0.00108841, 0.00206625]
Errors: [0.000341398, 2.24797e-5, 2.25417e-6, 3.20031e-7]

Name: Rodas3
Times: [0.000445113, 0.000668269, 0.00110876, 0.00224957]
Errors: [0.000937317, 6.24324e-6, 1.22523e-6, 1.59544e-7]

Name: TRBDF2
Times: [0.000398845, 0.000547485, 0.000754979, 0.00121249]
Errors: [0.000108409, 7.92059e-5, 1.6268e-5, 9.32119e-6]

Name: CVODE_BDF
Times: [0.000428983, 0.000419278, 0.000532028, 0.000616659]
Errors: [0.00049696, 0.000104618, 1.38678e-5, 6.44557e-7]

Name: rodas
Times: [0.000559033, 0.000580169, 0.000707072, 0.000927858]
Errors: [0.00369423, 1.48695e-6, 5.04545e-8, 1.63525e-8]

Name: radau
Times: [0.000763001, 0.00083118, 0.000963938, 0.00132802]
Errors: [3.49204e-6, 8.06599e-7, 1.4645e-7, 1.14391e-9]

Name: lsoda
Times: [0.000164521, 0.000222644, 0.000302167, 0.000354788]
Errors: [0.000374798, 6.99503e-5, 5.16518e-6, 5.80184e-7]

# v0.6

Name: OrdinaryDiffEq.Rosenbrock23
Times: [0.000272527, 0.000421401, 0.000743899, 0.0014318]
Errors: [0.000341398, 2.24797e-5, 2.25417e-6, 3.20031e-7]

Name: OrdinaryDiffEq.Rodas3
Times: [0.000438248, 0.000641923, 0.00115128, 0.00228896]
Errors: [0.000937317, 6.24324e-6, 1.22523e-6, 1.59544e-7]

Name: OrdinaryDiffEq.TRBDF2
Times: [0.000378822, 0.000439127, 0.000604467, 0.000940533]
Errors: [0.000202253, 7.92059e-5, 1.6268e-5, 9.32119e-6]

Name: Sundials.CVODE_BDF
Times: [0.000298933, 0.000330022, 0.000466527, 0.000515751]
Errors: [0.00049696, 0.000104618, 1.38678e-5, 6.44557e-7]

Name: ODEInterfaceDiffEq.rodas
Times: [0.000621021, 0.000663761, 0.000891177, 0.0011801]
Errors: [0.00369423, 1.48695e-6, 5.04545e-8, 1.63525e-8]

Name: ODEInterfaceDiffEq.radau
Times: [0.00090195, 0.000948701, 0.00115025, 0.0015165]
Errors: [3.49204e-6, 8.06599e-7, 1.4645e-7, 1.14389e-9]

Name: LSODA.lsoda
Times: [0.000242243, 0.000305563, 0.0003928, 0.000432101]
Errors: [0.000374798, 6.99503e-5, 5.16518e-6, 5.80184e-7]

Non-OrdinaryDiffEq.jl algorithms got better except for Sundials. OrdinaryDiffEq.jl algorithms got worse, moreso for more broadcast