Some changes necessary for subsurface

[Leguark]

From dict method

• I have added a method to be able to pass directly a python dictionary. It is just a simplification of the method from_csv. Since I created the pull request someone else has created another from_dict method. I just renamed my contribution to from_dict_advance since it does more post processing.

I am using this in subsurface to be able to first read a table with pandas, clean it and then converting the pandas dataframe to a valid dictionary that striplog understands.

Supporting integer as formation name

• At the moment any formation name that is passed as a integer is directly converted to float. This gave problems when the user uses ids for the formations and we want to map it to a table of the components.
• In the same direction, comparing two Component objects with different integers was always comparing them as True.

This is the test that needs all of the mention here: https://github.com/softwareunderground/subsurface/blob/6543c9924dcfd5d67068069f2a6fc1f9e17f5ef5/tests/test_io/test_welly_to_subsurface.py#L420

[coveralls]

Coverage decreased (-0.2%) to 76.815% when pulling 2d1b718 on Leguark:master into 02ae02e on agile-geoscience:master.

[coveralls]

Coverage decreased (-0.3%) to 75.11% when pulling fcd7825 on Leguark:master into 0c68f63 on agile-geoscience:master.

[ThirstyGeo]

@Leguark I'm just getting to know the striplog and Markov Chain object preferences and wondering if you can suggest how you'd clean up a log in the form of a pandas table with well, top, base, log. Based on @kwinkunks notebook it looks like a .GroupBy or .to_json might be the way to go.

[ThirstyGeo]

Something like this, I suppose:
df.dropna(how='any').groupby('well')['log'].apply(lambda x: x.to_json(orient='records'))