Improve and fix *Spline docstrings and code

[erheron]

During the investigation of the issue #9718 I've discovered a few possibilities for both improvements and fixes in the documentation of different *Spline classes in
https://github.com/scipy/scipy/blob/master/scipy/interpolate/fitpack2.py
I create this issue mainly because it seems to me that it'll be (and must be) more general than #9718
The things below will be mostly about RectBivariateSpline class

Things that came into mind:

• In BivariateSpline class, as far as I understand, RectBivariateSpline should be added to the end of the sentence, because it subclasses BivariateSpline
  

  scipy/scipy/interpolate/fitpack2.py

  Lines 931 to 933 in b818978

  	This class is meant to be subclassed, not instantiated directly.
  	To construct these splines, call either `SmoothBivariateSpline` or
  	`LSQBivariateSpline`.

• Docstring of RectBivariateSpline containes a few more inconsistencies: e.g. tx, ty in here:

  scipy/scipy/interpolate/fitpack2.py

  Lines 1149 to 1152 in b818978

  	bbox : array_like, optional
  	Sequence of length 4 specifying the boundary of the rectangular
  	approximation domain. By default,
  	``bbox=[min(x,tx),max(x,tx), min(y,ty),max(y,ty)]``.

  
  although they do not appear in init:
  

  scipy/scipy/interpolate/fitpack2.py

  Line 1169 in b818978

  def __init__(self, x, y, z, bbox=[None] * 4, kx=3, ky=3, s=0):

• I have a feeling that "See Also" sections in different Spline's might be more... broad? Consistent? There's not so many of that classes, so why not to mention them all, or at least ensure that if A -> B and A -> C (-> means subclassing), then B mentions C in "See Also" and vice versa?
• I've got a feeling that two last checks of these 4 are redundant, but I'm not sure
  

  scipy/scipy/interpolate/fitpack2.py

  Lines 1171 to 1178 in b818978

  	if not np.all(diff(x) > 0.0):
  	raise ValueError('x must be strictly increasing')
  	if not np.all(diff(y) > 0.0):
  	raise ValueError('y must be strictly increasing')
  	if not ((x.min() == x[0]) and (x.max() == x[-1])):
  	raise ValueError('x must be strictly ascending')
  	if not ((y.min() == y[0]) and (y.max() == y[-1])):
  	raise ValueError('y must be strictly ascending')

• Problem with DOC: docstring inconsistency in RectBivariateSpline #9718 remains for now, but I'm working on it

If this issue "makes sense" to some of the members, I suggest closing #9718 in favor of this one.

[stefanv]

Thank you, @erheron, a thorough review of this and related docs would be great. All the inconsistencies you noted should definitely be addressed.