Align parameters and hyperparameters with planned changes in scikit-learn

[desilinguist]

This PR makes the following changes in anticipation of the changes that will be coming in scikit-learn v0.22 (#438). We are making them now since v2.0 is a major release and we can pack in backwards incompatible changes here rather than later:

• Change the default value of gamma to scale for SVR and SVC and add scale as a value to the default grids.
• Change the default values of max_iter and tol to match the values for SGDClassifier and SGDRegressor.
• Change the value of max_iter for LogisticRegression to be 1000 since the current default of 100 on the scikit-learn side has been acknowledged as a major bug.
• Update test data due to changes in parameters.
• Some minor cleanup.
• Also link the RTD contributions page from the contribution guidelines in the repo (#526) since it was just a one line change.

[codecov[bot]]

Codecov Report

Merging #533 into master will increase coverage by <.01%. The diff coverage is 100%.

@@            Coverage Diff            @@
##           master    #533      +/-   ##
=========================================
+ Coverage    94.1%   94.1%   +<.01%     
=========================================
  Files          20      20              
  Lines        2918    2919       +1     
=========================================
+ Hits         2746    2747       +1     
  Misses        172     172

Impacted Files	Coverage Δ
skll/config.py	95.12% <ø> (ø)	:arrow_up:
skll/learner.py	93.67% <100%> (ø)	:arrow_up:

---

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 114d088...0b43a0a. Read the comment docs.

[desilinguist]

@aoifecahill I'd really like your review for this PR as well. Thanks!