Fix, improve and unify all unit tests

[omlins]

Some unit tests were not executed.

Unit test launch status before fix (launched/avail for Threads, for CUDA):
ParallelKernel:

1. allocators: (3/3, 3/3)
2. pk-init: (12/12, 12/12)
3. pk-reset: (7/7, 7/7)
4. hide_comm: (6/6, 31/31)
5. pk-parallel: (25/30, 25/33)
6. kernel_language (4/30, 4/12)

ParallelStencil:

1. ps-init (16/16, 16/16)
2. ps-reset (9/9, 9/9)
3. incremental comp (1/1, 1/1)
4. FD 1-D (14/14, 14/14)
5. FD 2-D (26/26, 26/26)
6. FD 3-D (41/41, 41/41)
7. ps-parallel (11/16, 11/17)

Unit test launch status after fix (launched/avail for Threads, for CUDA):

ParallelKernel:

1. allocators: (3/3, 3/3)
2. pk-init: (12/12, 12/12)
3. pk-reset: (7/7, 7/7)
4. hide_comm: (6/6, 31/31)
5. pk-parallel: (30/30, 33/33)
6. kernel_language (30/30, 12/12)

ParallelStencil:

1. ps-init (16/16, 16/16)
2. ps-reset (9/9, 9/9)
3. incremental comp (1/1, 1/1)
4. FD 1-D (14/14, 14/14)
5. FD 2-D (26/26, 26/26)
6. FD 3-D (41/41, 41/41)
7. ps-parallel (16/16, 17/17)