Add sequences for ImplicitEulerExtrapolation

[utkarsh530]

Hi,

I added a choice of sequences in ImplicitEulerExtrapolation similar to seulex. The default sequence is set to Bulirsch. There we some minor improvements in the method. Comparing with seulex:

function test!(du, u, p, t)
     du[1] = -10 * u[1]
     du[2] = -0.01 * u[2]
 end
 u0 = [1.01;1.01]
 tspan = (0.0, 1.0)
 prob = ODEProblem(test!, u0, tspan)

julia> sol = solve(prob,seulex())
retcode: Success
Interpolation: 1st order linear
t: 12-element Array{Float64,1}:
 0.0
 1.0e-6
 7.957010852370433e-6
 5.6357010852370424e-5
 ⋮
 0.171697546989057
 0.362553477748044
 0.8757951760869469
 1.0
u: 12-element Array{Array{Float64,1},1}:
 [1.01, 1.01]
 [1.0099899000505004, 1.0099999898999994]
 [1.009919637387684, 1.0099999196341929]
 [1.0094309545603295, 1.0099994307943505]
 ⋮
 [0.18151135597475912, 1.0082673426661455]
 [0.02691950277797022, 1.006344839833677]
 [0.00015776855888100059, 1.001193090258896]
 [4.575739308257184e-5, 0.9999503320867008]

julia> sol = solve(prob,ImplicitEulerExtrapolation(),dt=1e-6)
┌ Warning: Threading in `ImplicitEulerExtrapolation` is currently disabled. Thus `threading` has been changed from `true` to `false`.
└ @ OrdinaryDiffEq C:\Users\Utkarsh\.julia\dev\OrdinaryDiffEq\src\algorithms.jl:65
retcode: Success
Interpolation: 3rd order Hermite
t: 14-element Array{Float64,1}:
 0.0
 1.0e-6
 1.1e-5
 0.00011099999999999999
 ⋮
 0.33900999182564423
 0.5605959154883677
 0.8354148791156211
 1.0
u: 14-element Array{Array{Float64,1},1}:
 [1.01, 1.01]
 [1.0099899000505, 1.0099999899]
 [1.0098889061104441, 1.009999888900006]
 [1.0088795221484086, 1.0099988789006225]
 ⋮
 [0.03404236668264801, 1.006581796382659]
 [0.0037126518729961797, 1.00435382215883]
 [0.00023777180592478287, 1.0015974566391057]
 [4.585389370221542e-5, 0.999950332087402]

I will make changes on adaptive step-sizing (which seems to be different from Hairer II for seulex) after this. It would help me compare benchmarks across different sequences.
@ChrisRackauckas please review.

[ChrisRackauckas]

Looks correct to me, but it should get convergence tests for the different sequence choices. See how the other convergence tests set the minimum and maximum order.