Fixes devices vector alloc to fix seg fault, removes unused RAFT code in PLC, re-enables full CI testing

[rlratzel]

closes #3124

• Adds check to avoid allocating and copying zero-length device vectors. This prevents the seg fault shown below.
• Removes the special case to ignore seg faults in CI scripts
• Adds a test to reproduce seg fault locally (see output below).

This PR addresses the problem shown below:

================================= test session starts =================================
platform linux -- Python 3.8.15, pytest-7.2.0, pluggy-1.0.0 -- /opt/conda/envs/test/bin/python3.8
cachedir: .pytest_cache
rapids_pytest_benchmark: 0.0.14
benchmark: 3.2.3 (defaults: timer=time.perf_counter disable_gc=False min_rounds=5 min_time=0.000005 max_time=1.0 calibration_precision=10 warmup=False warmup_iterations=100000)
rootdir: /cugraph/python/pylibcugraph, configfile: pytest.ini
plugins: cov-4.0.0, rapids-pytest-benchmark-0.0.14, benchmark-3.2.3
collected 9 items / 8 deselected / 1 selected

python/pylibcugraph/pylibcugraph/tests/test_graph_sg.py::test_SGGraph_create_from_cudf
get edgelist...edgelist =     src  dst  wgt
0    0    1  0.0
1    1    2  0.1
2    2    4  0.2
done
create Graph...done
created SGGraph plc_graph=<pylibcugraph.graphs.SGGraph object at 0x7fb7e35f30f0>
PASSED

=========================== 1 passed, 8 deselected in 1.69s ===========================
Segmentation fault (core dumped)

@cjnolet found a work-around for us, so this should pass CI and can be merged after rapidsai/raft#1224

[codecov-commenter]

Codecov Report

Base: 56.26% // Head: 56.26% // No change to project coverage 👍

Coverage data is based on head (3bba486) compared to base (64dabc3).
Patch has no changes to coverable lines.

Additional details and impacted files

@@              Coverage Diff              @@
##           branch-23.02    #3167   +/-   ##
=============================================
  Coverage         56.26%   56.26%           
=============================================
  Files               153      153           
  Lines              9658     9658           
=============================================
  Hits               5434     5434           
  Misses             4224     4224           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[rlratzel]

I can't request changes on my own PRs, so I added the DO NOT MERGE label so this isn't accidentally merged before commenting on if removing RAFT static linking still needs to/can be done: #3167 (comment)

[rlratzel]

Added DO NOT MERGE back until I've finished verifying it's safe to remove the RAFT sources from PLC, in the off chance CI passes even if they're needed.

[rlratzel]

/merge

[ChuckHastings]

Cancelled CI run. We need rapidsai/raft#1224 to be pushed into nightlies so that the change can be picked up by CI.

[ChuckHastings]

/merge