add ITP bracketing method

[jverzani]

cf discourse

To match Wikipedia's example:

julia> solve(ZeroProblem(x -> x^3 - x - 2, (1,2)), 
       Roots.ITP(κ₁=0.1, κ₂=2, n₀=1), 
       xatol=1e-5, xrtol=0.0, 
       verbose=true)
Results of univariate zero finding:

* Converged to: 1.5213830127326777
* Algorithm: Roots.ITP{Float64, Int64}(0.1, 2, 1)
* iterations: 6
* function evaluations ≈ 7
* stopped as x_n ≈ x_{n-1} using atol=xatol, rtol=xrtol

Trace:
(a_0, b_0) = ( 0.0000000000000000,  2.0000000000000000)
(a_1, b_1) = ( 1.0000000000000000,  2.0000000000000000)
(a_2, b_2) = ( 1.4333333333333333,  2.0000000000000000)
(a_3, b_3) = ( 1.4333333333333333,  1.5271314505696607)
(a_4, b_4) = ( 1.5200928115097818,  1.5271314505696607)
(a_5, b_5) = ( 1.5213789911605158,  1.5271314505696607)
(a_6, b_6) = ( 1.5213789911605158,  1.5213830127326777)


1.5213789911605158

[codecov]

Codecov Report

Merging #274 (80cc4f0) into master (3956382) will increase coverage by 0.33%.
The diff coverage is 98.14%.

@@            Coverage Diff             @@
##           master     #274      +/-   ##
==========================================
+ Coverage   83.82%   84.15%   +0.33%     
==========================================
  Files          12       12              
  Lines        2015     2064      +49     
==========================================
+ Hits         1689     1737      +48     
- Misses        326      327       +1     

Impacted Files	Coverage Δ
src/bracketing.jl	87.98% <98.14%> (+1.04%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9128907...80cc4f0. Read the comment docs.

[oscardssmith]

On 1.8 this will be kind of fast if k_2 is an integer, but hard-coding k_2 to 2 would still be faster.

[KronosTheLate]

Thank you for putting together this PR so quickly - it looks great, and I am thrilled to have my ITP implementation in such an established package <3