Issue with single-dimensional variables

Hello, this may have been reported before but there is an annoying issue with size one variables when trying to multiply them by a matrix.

Here is an example. I am encoding the constraints `1 <= x <= 2` as `x <= 2` and `-x <= -1`.
```python
from cylp.cy import CyClpSimplex
from cylp.py.modeling.CyLPModel import CyLPArray
import numpy as np

s = CyClpSimplex()
x = s.addVariable('x', 1)
A = np.matrix([[1, -1]])
b = np.array([2, -1])
s.addConstraint(A*x <= b)
s.objective = x.sum()
s.primal()
```

This example crashes with the following error message:
```
IndexError Traceback (most recent call last)
in
7 A = np.matrix([[1, -1]])
8 b = np.array([2, -1])
----> 9 s.addConstraint(A*x <= b)
10 s.objective = x.sum()
11 s.primal()

cylp\cy\CyClpSimplex.pyx in cylp.cy.CyClpSimplex.CyClpSimplex.addConstraint()

C:\ProgramData\Anaconda3\lib\site-packages\cylp\py\modeling\CyLPModel.py in addConstraint(self, cons, consName, addMpsNames)
1016 '''
1017
-> 1018 c = cons.evaluate(consName)
1019 if not c.isRange:
1020 self.constraints.append(c)

C:\ProgramData\Anaconda3\lib\site-packages\cylp\py\modeling\CyLPModel.py in evaluate(self, name)
291 right = operands.pop()
292 left = operands.pop()
--> 293 cons.perform(token, left, right)
294 operands.append(CyLPExpr(token, left, right))
295

C:\ProgramData\Anaconda3\lib\site-packages\cylp\py\modeling\CyLPModel.py in perform(self, opr, left, right)
402 raise Exception("Coefficient:\n%s\n has %d" \
403 " columns expected %d"
--> 404 % (left, left.shape[1], right.dim))
405 if len(left.shape) == 1: # Array
406 nr = 1

IndexError: tuple index out of range
```

I think the issue comes from `np.squeeze`ing the matrix at the beginning of `CyLPConstraint.perform` ([link](https://github.com/coin-or/CyLP/blob/master/cylp/py/modeling/CyLPModel.py#L358)) which transforms the matrix to an array when one of the dimensions is 1.
I don't know if this can be fixed easily?
I ran into this issue when solving over multiple datasets with a loop: one of the datasets only had one number and created a size 1 variable.
I managed to work around the issue by adding an extra variable and padding my matrices with zeros. A bit annoying and ugly but it did the trick.