Clean up stock sample data.

[QuLogic]

This is a followup to #7559 with the cleanup of the stock sample data that was taken out of there.

AAPL and INTC are no longer required without stock_demo.py, plus there are 3 copies of AAPL data. Convert datetime.date to np.datetime64 to increase portability of the files.

While using np.datetime64 in the NumPy files is more portable across Python versions, Matplotlib still really only supports datetime objects, so I've had to convert back after loading. This is a bit suboptimal (but not really worse than the existing setup), so now that we depend on NumPy 1.7 which has had a few improvements, it may be time to look into importing the np.datetime64 converter.

[Kojoley]

Disclaimer: I've checked examples only on Py3.5

[codecov-io]

Current coverage is 62.16% (diff: 100%)

Merging #7645 into master will increase coverage by 0.09%

@@             master      #7645   diff @@
==========================================
  Files           174        174          
  Lines         56021      56664   +643   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits          34773      35225   +452   
- Misses        21248      21439   +191   
  Partials          0          0          

Powered by Codecov. Last update ab98852...9ba1fd5

[NelleV]

Thanks for the clean up!
This looks good.