DOC: Refactor numeric index docs

[gfyoung]

Refactor NumericIndex docs to avoid duplicate documentation. xref #14937.

[codecov-io]

Current coverage is 84.78% (diff: 100%)

Merging #14985 into master will increase coverage by <.01%

@@             master     #14985   diff @@
==========================================
  Files           145        145          
  Lines         51090      51107    +17   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits          43315      43333    +18   
+ Misses         7775       7774     -1   
  Partials          0          0          

Powered by Codecov. Last update 7f0eefc...ba80766

[jreback]

nit: needs a blank like before Paramaters. need to wrap the line for Int64Index.

In [1]: pd.Int64Index?
Init signature: pd.Int64Index(data=None, dtype=None, copy=False, name=None, fastpath=False)
Docstring:     
Immutable ndarray implementing an ordered, sliceable set. The basic object
storing axis labels for all pandas objects. Int64Index is a special case
of `Index` with purely integer labels. This is the default index type used by the DataFrame and Series ctors when no explicit index is provided by the user.
Parameters
----------
data : array-like (1-dimensional)
dtype : NumPy dtype (default: int64)
copy : bool
    Make a copy of input ndarray
name : object
    Name to be stored in the index
Notes
-----
An Index instance can **only** contain hashable objects.
File:           ~/pandas/pandas/indexes/numeric.py
Type:           type

In [2]: pd.Float64Index?
Init signature: pd.Float64Index(data=None, dtype=None, copy=False, name=None, fastpath=False)
Docstring:     
Immutable ndarray implementing an ordered, sliceable set. The basic object
storing axis labels for all pandas objects. Float64Index is a special case
of `Index` with purely integer labels. 
Parameters
----------
data : array-like (1-dimensional)
dtype : NumPy dtype (default: float64)
copy : bool
    Make a copy of input ndarray
name : object
    Name to be stored in the index
Notes
-----
An Index instance can **only** contain hashable objects.
File:           ~/pandas/pandas/indexes/numeric.py
Type:           type

[gfyoung]

@jreback : Made the requested doc changes. Ready to merge if there are no other concerns.

[jorisvandenbossche]

+1 on sharing docstrings for public functions (eg the classes in this PR), but I am not sure we should also do this for private helper functions.
As the docstrings there are in the first place helpful for developers when looking and trying to understand the code. That the docstring is right there next to the function when looking at it in your IDE can be very helpful.

[jorisvandenbossche]

also a blank line missing above 'Parameters' here I think

[gfyoung]

Good catch. Done.

[jorisvandenbossche]

"purely integer labels" -> will also be used for Float64Index?

[gfyoung]

Good catch. Done.

[gfyoung]

@jorisvandenbossche : I think sharing docstrings across private functions is fine. I get more confused by the inconsistencies across written-out docstrings for the same functions, whether private or public. Not to mention, there are already numerous private functions in base.py that docstring-shared.

[gfyoung]

@jreback , @jorisvandenbossche : If there are no other concerns, I think this is good to merge.

[jreback]

thanks!