include callback, now we can save .csv each observation

[rspadim]

from kaggle https://www.kaggle.com/c/home-credit-default-risk/discussion/59195#345409

include callback is important to save csv from each step, since xgb and others algoritms can crash with memory limits and lost work

[codecov-io]

Codecov Report

Merging #100 into master will decrease coverage by 1.05%.
The diff coverage is 33.33%.

@@            Coverage Diff             @@
##           master     #100      +/-   ##
==========================================
- Coverage   82.87%   81.81%   -1.06%     
==========================================
  Files           4        4              
  Lines         292      297       +5     
  Branches       35       37       +2     
==========================================
+ Hits          242      243       +1     
- Misses         45       47       +2     
- Partials        5        7       +2

Impacted Files	Coverage Δ
bayes_opt/bayesian_optimization.py	64.16% <33.33%> (-1.93%)	⬇️

---

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 317f735...87f6693. Read the comment docs.

[fmfn]

I really like the idea of implementing callbacks, however I think it needs a bit more. For instance which parameters are passed to callbacks. Making callbacks an object you can build on top of. And making sure callbacks are called at the end of every step (and not only when initializing).

[fmfn]

We now have the concept of observables in the library, and with the next release (under way) I intend to include this functionality. Stay tuned.