Fix API mismatch in Tensorflow Optimizers

[lazyoracle]

What

Update tf_sgd to use maxiters and include note on WIP TF Optimizers

Why

Closes #102 and associates

How

• Add warnings to the adam, adadelta and rmsprop optimizers. These have incomplete/incorrect implementations while some of the default hyper-parameters have been altered without adequate reasoning or comments. Pending further investigations into these algorithms or without the option to alter these hyper-parameters from the user-space, these are essentially buggy and must be avoided. Also added # TODOs to make note of what needs to be done
• Updated the tf_sgd optimizer to use the correct maxiters parameter instead of maxfun. This is because these optimizers allow for directly controlling the number of algorithm iterations and not function evaluations.
• Added an Error Handling code in case someone passes maxfun to this optimizer
• Provided an option for setting the learning rate and momentum from the user-space
• Added and updated tests to also check for error handling code

[codecov]

Codecov Report

Merging #104 (2fc4436) into dev (b6dd98f) will increase coverage by 0.02%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##              dev     #104      +/-   ##
==========================================
+ Coverage   57.32%   57.34%   +0.02%     
==========================================
  Files          35       35              
  Lines        5279     5291      +12     
==========================================
+ Hits         3026     3034       +8     
- Misses       2253     2257       +4     

Impacted Files	Coverage Δ
c3/libraries/algorithms.py	19.23% <66.66%> (+1.71%)	⬆️
c3/optimizers/c1.py	73.73% <0.00%> (+1.01%)	⬆️

---

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 b6dd98f...2fc4436. Read the comment docs.