Implement relational operations <=, ==, >=, <, >

[certik]

For example, this:

 auto formula = parse("((x>=y) & (y>=z))*sign(x-y)")

Would work like the following:

 result = formula.lambda(x=3, y=2.5, z=1.5) = ((1.0)&(1.0))*1.0 = 1.0
 result = formula.lambda(x=3, y=3.5, z=1.5) = ((0.0)&(1.0))*1.0 = 0.0

In SymPy, you can do things like:

In [1]: x <= y
Out[1]: x ≤ y

[isuruf]

You can do this with Piecewise
Piecewise( ((x-y in [0, inf)) and (y-z in [0, inf)), (x-y)/abs(x-y)), (True, 0))

[bjodah]

FWIW, relationals in symengine would help a use-case I just encountered for Lambdify:

>>> cb = symengine.Lambdify([x, y], [x>=0, y>=0])
...
SympifyError: sympy2symengine: Cannot convert 'x >= 0' to a symengine type.

Where I would like to have this eventually:

>>> np.allclose(cb([2, -1]), [1.0, 0.0])
True

[isuruf]

See symengine/symengine.py#90. In it I used intervals (which was already implemented) and contains. to_contains can convert x >= 0 to a symengine expression.
Now that And, Or, Not are all there, after implementing relationals, we don't need contains at all.

[ShikharJ]

Fixed via #1276.