Cleaning FloatArrayFormatter

[nbonnotte]

While working on issue #12031 (adding a decimal parameter to .to_latex() and .to_html()), I realized a solution would be pretty straight-forward if only FloatArrayFormatter was a bit cleaned.

Currently, there are two methods that do a similar work:

• .get_formatted_data(), which returns the array with its float values converted to strings using the parameters given at initialisation.
• .get_result(), which does something similar, but with a fixed-width output suitable for screen printing.

The two methods don't share any piece of code.

(The first method can already deal with a decimal parameter. The second cannot, and it is the one used by .to_latex.)

This is in part my fault, since my pull request #11681 could have been more factorized.

Anyway, I'm volunteering to do a PR.

[jreback]

sure!

[jreback]

related this this is:

In [1]: s = Series([1,2,3],np.arange(3.))

In [2]: s
Out[2]: 
0    1
1    2
2    3
dtype: int64

In [3]: s.index
Out[3]: Float64Index([0.0, 1.0, 2.0], dtype='float64')

In [4]: s = Series([1,2,3],np.arange(3.)+.01)

In [5]: s
Out[5]: 
0.01    1
1.01    2
2.01    3
dtype: int64

In [6]: s.index
Out[6]: Float64Index([0.01, 1.01, 2.01], dtype='float64')

This is misleading and [2] index should prob be format with a single decimal place
eg
1.0 to make it obvious this is a Float64Index.

the _trim_zeros function is a bit aggressive here.

[nbonnotte]

@jreback I'm about to make a separate PR for the _trim_zeros issue