Reproduce two minima in _make_cross_face_batch Gauss-Newton

[inducer]

This is a refreshed version of the reproducer for two minima in the Gauss-Newton-based inversion of the reference-to-global map in _make_cross_face_batch described in https://github.com/inducer/meshmode/pull/105#issuecomment-862003829.

To reproduce it, clone this branch and run

 pycl test_connection.py 'test_bdry_restriction_is_permutation(_acf, PolynomialEquidistantSimplexGroupFactory, 3, 2)'  

The output currently is

GOOD UNIT NODES
[[-1. -1. -1.  0.  0.  1.]
 [-1.  0.  1. -1.  0. -1.]]
BAD UNIT NODES
[[-1.00000000e+00 -1.00000000e+00 -9.86265258e-01 -5.73886430e-16
  -5.09506980e-16  1.00000000e+00]
 [-1.00000000e+00 -1.23179577e-16  1.00000000e+00 -1.00000000e+00
  -1.61968491e-15 -1.00000000e+00]]

FIXED RESIDUAL
1.3322676295501878e-15
UNFIXED RESIDUAL
1.3322676295501878e-15