As explained in NEP 32, the financial functions are to be removed
from version 1.20.
They are now replaced with module level `__getattr__` to give a useful
error message for those surprised by the `AttributeError`.

This only works for Python 3.7+, but it is expected that by the 1.20 release
Python 3.6 will not be supported.

Add closed option to randint to simplify some cases

skip doctests that require scipy
move original mtrand module to _mtrand
adjust documentation for namespace change

first cut at building randomgen
upgrade 'cythonize' and fix absolute imports to relative
define NPY_NO_DEPRECATED_API and fix other warnings
enable pgc64 by always using PCG_EMULATED_MATH
refactor so import randomgen works
add TODO comments for pcg64 improvements
fix imports, module name in setup.py; remove _testing
make cythonize non-recursive, restore examples to proper place
update to randomgen 7bca296c0b9
replace mtrand with LegacyGenerator, tweak for compatibility
port f879ef4 to fix GH10839
minimized difference between generator.pyx and _legacy.pyx
fix namespace in doctests, mark results that are random
update to randomgen commit 95c8cdd1c
Incorporate testing of edge cases into main tests
Rename test files to describe their purpose
Import import locations to reflect numpy paths
Correct tolerance on float32 tests
Remove set_printoptions
Remove complex normal
Remove future imports
Pull in BasicRNG source changes from original author
Small doc fixes
_mtrand => _rand
Improve consistency of nan handling
Prevent nans prducing values from int functions
add randomgen documentation to the tree

The user guide was missing a quick tutorial --- the basics.* stuff is
somewhat too complex already.

The "building numpy" instructions also should not be "introductory
material".

Also fix a duplicate entry for numpy.testing due to incorrect
header underlining.

The original masked-NA-NEP branch contained a large number of changes
in addition to the core NA support. For example:
 - ufunc.__call__ support for where= argument
 - nditer support for arbitrary masks (in support of where=)
 - ufunc.reduce support for simultaneous reduction over multiple axes
 - a new "array assignment API"
 - ndarray.diagonal() returning a view in all cases
 - bug-fixes in __array_priority__ handling
 - datetime test changes
etc. There's no consensus yet on what should be done with the
maskna-related part of this branch, but the rest is generally useful
and uncontroversial, so the goal of this branch is to identify exactly
which code changes are involved in maskna support.

The basic strategy used to create this patch was:
 - Remove the new masking-related fields from ndarray, so no arrays
   are masked
 - Go through and remove all the code that this makes
   dead/inaccessible/irrelevant, in a largely mechanical fashion. So
   for example, if I saw 'if (PyArray_HASMASK(a)) { ... }' then that
   whole block was obviously just dead code if no arrays have masks,
   and I removed it. Likewise for function arguments like skipna that
   are useless if there aren't any NAs to skip.

This changed the signature of a number of functions that were newly
exposed in the numpy public API. I've removed all such functions from
the public API, since releasing them with the NA-less signature in 1.7
would create pointless compatibility hassles later if and when we add
back the NA-related functionality. Most such functions are removed by
this commit; the exception is PyArray_ReduceWrapper, which requires
more extensive surgery, and will be handled in followup commits.

I also removed the new ndarray.setasflat method. Reason: a comment
noted that the only reason this was added was to allow easier testing
of one branch of PyArray_CopyAsFlat. That branch is now the main
branch, so that isn't an issue. Nonetheless this function is arguably
useful, so perhaps it should have remained, but I judged that since
numpy's API is already hairier than we would like, it's not a good
idea to add extra hair "just in case". (Also AFAICT the test for this
method in test_maskna was actually incorrect, as noted here:
   https://github.com/njsmith/numpyNEP/blob/master/numpyNEP.py
so I'm not confident that it ever worked in master, though I haven't
had a chance to follow-up on this.)

I also removed numpy.count_reduce_items, since without skipna it
became trivial.

I believe that these are the only exceptions to the "remove dead code"
strategy.

The various padding functions are exposed as options to a public 'pad'
function.  Example:

    pad(a, 5, mode='mean')

Current modes are 'constant', 'edge', 'linear_ramp', 'maximum', 'mean',
'median', 'minimum', 'reflect', 'symmetric', 'wrap', and <function>

This commit includes unit tests and doctests and is based on feature
request ticket #655.

Also:
1. Note that the polynomial package is preferred over poly1d.

2. Remove deprecation of mapparms in __init__.py as it interferes with the
   documention of the method of the same name. This is probably safe as it
   is unlikely to be used yet.

3. Make some improvements to the documentation in polytemplate.