I found some testcases, where the Codewars environment shows different results than my program:

number of walls: 6
walls = (
[ (Point(6, 2),Point(7, 1))
, (Point(7, 4),Point(6, 3))
, (Point(1, 2),Point(3, 0))
, (Point(2, 1),Point(0, 3))
, (Point(9, 1),Point(8, 2))
, (Point(6, 4),Point(7, 5))
])
width = 10
height = 5
expected = False
Your answer: True

Here the 2 overlapping edges (Point(1, 2),Point(3, 0)) and (Point(2, 1),Point(0, 3)) form a partition with upper and left boundary walls.

Another testcase:

number of walls: 11
walls = (
[ ((22, 7),(19, 5))
, ((21, 6),(21, 4))
, ((24, 7),(24, 6))
, ((18, 8),(18, 8))
, ((17, 8),(17, 8))
, ((23, 8),(24, 9))
, ((17, 9),(17, 7))
, ((19, 8),(16, 8))
, ((16, 9),(14, 9))
, ((20, 9),(17, 7))
, ((18, 9),(17, 6))
])
width = 40
height = 16
expected = False
Your answer: True

There are 2 partitions in the center.

C translation

hint link is broken in description.

c#: Times out just reading in the results from a switch-case statement.

It has confirmed the results as "solved" once before, but instead of submitting I cleaned up some comments and functions that were no longer used.

Python translation with 500 tests instead of 36736.

JS Translation

expected and actual are swapped in full tests suite.

JS mocha + chai framework should be used

C#: method name should be PascalCase (Please refer to implementation of backward compatibility here )

D translation

Rust translation

Java: random tests are inconsistent (my solution passes them after a few attempts).
The fault is the same: expected:<[]> but was:<[Z]>

20 random tests is not enough, after couple submission I passed with algo finding some suboptimal solutions (1 or 2 cases most of the time).