BUG: Sparse creation with object dtype may raise TypeError

[sinhrks]

closes #11633
closes #11856

Added basic tests for groupby and pivot_table.

[jreback]

hmm, maybe this should be more integrated with _sanitize_values (or that should be enhanced). not a big deal, but feel free to reorg code to make more general / readable code.

[sinhrks]

Can I leave this ATM, because this can be removed entirely after changing dtype default to None (#667).

[jreback]

yes - was just bringing it up mainly for future / FYI

[codecov-io]

Current coverage is 84.15%

Merging #13201 into master will increase coverage by <.01%

@@             master     #13201   diff @@
==========================================
  Files           138        138          
  Lines         50388      50394     +6   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits          42386      42407    +21   
+ Misses         8002       7987    -15   
  Partials          0          0          

Powered by Codecov. Last updated by 4e4a7d9...443b47e

[jreback]

thanks @sinhrks very nice as usual

side-note (nothing to do with this PR I think, but should investigate at some point).

nosetests pandas/sparse
....
/Users/jreback/pandas/pandas/sparse/array.py:52: FutureWarning: comparison to `None` will result in an elementwise object comparison in the future.
  return _wrap_result(name, op(self.sp_values, other),