Add accuracy binary classification objective

[angela97lin]

Closes #294

(Updating base branch from #559)

[codecov]

Codecov Report

Merging #584 into improved_objectives will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@                 Coverage Diff                  @@
##           improved_objectives     #584   +/-   ##
====================================================
  Coverage                98.95%   98.96%           
====================================================
  Files                      132      133    +1     
  Lines                     4504     4532   +28     
====================================================
+ Hits                      4457     4485   +28     
  Misses                      47       47           

Impacted Files	Coverage Δ
evalml/objectives/__init__.py	100.00% <ø> (ø)
evalml/objectives/utils.py	94.44% <ø> (ø)
evalml/exceptions/__init__.py	100.00% <100.00%> (ø)
evalml/exceptions/exceptions.py	100.00% <100.00%> (ø)
evalml/objectives/standard_metrics.py	99.51% <100.00%> (+0.02%)	⬆️
evalml/tests/objective_tests/test_objectives.py	96.29% <100.00%> (ø)
...lml/tests/objective_tests/test_standard_metrics.py	100.00% <100.00%> (ø)

---

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 a88e69c...d8be7c8. Read the comment docs.

[dsherry]

Looks good! Please add unit test coverage. Actually, looking now, I see on master we don't have direct coverage of our objectives from standard_metrics.py. If that's correct, let's create a new unit test file or something and just cover this new metric, because it's important to have direct coverage for all our objectives.

Also let's freeze this until we've merged improved_objectives to master, which I hope will be today!

[angela97lin]

@dsherry Yeah, the way that the objectives have been covered via tests thus far is in their usage in AutoML. We might have talked about this before but I'm not 100% sure what unit test coverage would look like that's helpful / not just "testing that sklearn is doing what it claims to be doing," with the objective function?

[dsherry]

Yeah, pretty much! We need coverage of each objective so that a) if we alter the implementation of an objective to do something more complicated than call sklearn, we know it conforms to the same standard as before, and b) in the unlikely but not impossible event that sklearn ships a bug, or our usage of their code somehow produces a bug, we catch it!

A great example is #588 : we should have unit test coverage for precision and f1 which asserts that the edge case described there has the behavior we expect, and that the warning suppression works.

[angela97lin]

Closing in favor of #624