UMAP Segmentation Faults

Nobody has claimed this yet.

Hi,

I thought it might be best to just create a new issue, for this as the issue seems a little different to https://github.com/lmcinnes/umap/issues/421 which I originally commented this error on.

I've started getting a segfault when trying to play around with the numba threading layers (setting it `tbb`) in order to use UMAP with ProcessPoolExecutor. It happened very suddenly, and now consistently happens whenever I try to run UMAP inside a script, regardless of threading layer or if it is running inside a process pool.

The weird thing is that the seg fault does not occur if I just run UMAP inside of a python terminal, it only occurs when I run it via command line through a script.

The error looks like this on one set of data:
```
Fatal Python error: Segmentation fault

Current thread 0x00007f5c2431c700 (most recent call first):
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/umap/umap_.py", line 580 in fuzzy_simplicial_set
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/umap/umap_.py", line 2373 in fit
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/flight/rosella/embedding.py", line 405 in fit_transform
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/flight/rosella/rosella.py", line 248 in perform_binning
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/flight/flight.py", line 442 in bin
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/flight/flight.py", line 365 in main
File "/home/n10853499/.conda/envs/rosella-dev/bin/flight", line 8 in
```

And there is secondary error on another set of data that looks like this:

```
Fatal Python error: Segmentation fault

Thread 0x00007f22211a4700 (most recent call first):
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/pynndescent/pynndescent_.py", line 874 in __inFatal Python error: iSegmentation faultt

__
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/umap/umap_.py", line 328 in nearest_neighbors
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/umap/umap_.py", line 2415 in fit
File "/home/n10853499/.conda/envs/rosella-dev/lib/python3.8/site-packages/flight/rosella/embedding.py", line 405 in fit_transform
File "/home/n10853499/Segmentation fault (core dumped)
```

Downgrading numba doesn't help with this issue, nor does downgrading pynndescent or using the master branch from the pynndescent github. Additionally, this is happening on a fresh conda environment so something pretty odd seems to be happening.

and my conda environment looks like this:

```
# packages in environment at /home/n10853499/.conda/envs/rosella-dev:
#
# Name Version Build Channel
_libgcc_mutex 0.1 conda_forge conda-forge
_openmp_mutex 4.5 1_gnu conda-forge
attrs 21.2.0 pyhd8ed1ab_0 conda-forge
backcall 0.2.0 pyh9f0ad1d_0 conda-forge
backports 1.0 py_2 conda-forge
backports.functools_lru_cache 1.6.4 pyhd8ed1ab_0 conda-forge
biopython 1.79 py38h497a2fe_0 conda-forge
blis 0.8.1 h7f98852_1 conda-forge
brotlipy 0.7.0 py38h497a2fe_1001 conda-forge
bwa 0.7.17 h5bf99c6_8 bioconda
bzip2 1.0.8 h7f98852_4 conda-forge
ca-certificates 2021.5.30 ha878542_0 conda-forge
cachecontrol 0.12.6 py_0 conda-forge
certifi 2021.5.30 py38h578d9bd_0 conda-forge
cffi 1.14.4 py38ha312104_0 conda-forge
chardet 4.0.0 py38h578d9bd_1 conda-forge
charset-normalizer 2.0.0 pyhd8ed1ab_0 conda-forge
cryptography 3.4.7 py38ha5dfef3_0 conda-forge
curl 7.71.1 he644dc0_3 conda-forge
cycler 0.10.0 py_2 conda-forge
cython 0.29.24 py38h709712a_0 conda-forge
decorator 5.0.9 pyhd8ed1ab_0 conda-forge
flight-genome 1.2.1 pyh5e36f6f_0 bioconda
freetype 2.10.4 h0708190_1 conda-forge
gsl 2.6 he838d99_2 conda-forge
hdbscan 0.8.27 py38h5c078b8_0 conda-forge
hdmedians 0.14.2 py38hb5d20a5_0 conda-forge
htslib 1.9 h4da6232_3 bioconda
idna 3.1 pyhd3deb0d_0 conda-forge
imageio 2.9.0 py_0 conda-forge
iniconfig 1.1.1 pyh9f0ad1d_0 conda-forge
ipython 7.26.0 py38he5a9106_0 conda-forge
ipython_genutils 0.2.0 py_1 conda-forge
jedi 0.18.0 py38h578d9bd_2 conda-forge
joblib 0.17.0 py_0 conda-forge
jpeg 9d h36c2ea0_0 conda-forge
k8 0.2.5 h9a82719_1 bioconda
kiwisolver 1.3.1 py38h1fd1430_1 conda-forge
krb5 1.17.2 h926e7f8_0 conda-forge
lcms2 2.12 hddcbb42_0 conda-forge
ld_impl_linux-64 2.36.1 hea4e1c9_2 conda-forge
libcblas 3.9.0 10_openblas conda-forge [38/1790]
libcurl 7.71.1 hcdd3856_3 conda-forge
libdeflate 1.6 h516909a_0 conda-forge
libedit 3.1.20191231 h46ee950_2 conda-forge
libffi 3.2.1 he1b5a44_1007 conda-forge
libgcc-ng 11.1.0 hc902ee8_8 conda-forge
libgfortran-ng 11.1.0 h69a702a_8 conda-forge
libgfortran5 11.1.0 h6c583b3_8 conda-forge
libgomp 11.1.0 hc902ee8_8 conda-forge
liblapack 3.9.0 10_openblas conda-forge
libllvm10 10.0.1 he513fc3_3 conda-forge
libopenblas 0.3.17 pthreads_h8fe5266_1 conda-forge
libpng 1.6.37 h21135ba_2 conda-forge
libssh2 1.9.0 ha56f1ee_6 conda-forge
libstdcxx-ng 11.1.0 h56837e0_8 conda-forge
libtiff 4.3.0 hf544144_0 conda-forge
libwebp-base 1.2.0 h7f98852_2 conda-forge
llvmlite 0.36.0 py38h4630a5e_0 conda-forge
lockfile 0.12.2 py_1 conda-forge
lz4-c 1.9.3 h9c3ff4c_1 conda-forge
matplotlib-base 3.4.2 py38hcc49a3a_0 conda-forge
matplotlib-inline 0.1.2 pyhd8ed1ab_2 conda-forge
minimap2 2.21 h5bf99c6_0 bioconda
more-itertools 8.8.0 pyhd8ed1ab_0 conda-forge
msgpack-python 1.0.2 py38h1fd1430_1 conda-forge
natsort 7.1.1 pyhd8ed1ab_0 conda-forge
ncurses 6.1 hf484d3e_1002 conda-forge
numba 0.53.1 py38h8b71fd7_1 conda-forge
numpy 1.21.1 py38h9894fe3_0 conda-forge
olefile 0.46 pyh9f0ad1d_1 conda-forge
openblas 0.3.17 pthreads_h4748800_1 conda-forge
openjpeg 2.4.0 hb52868f_1 conda-forge
openssl 1.1.1k h7f98852_0 conda-forge
packaging 21.0 pyhd8ed1ab_0 conda-forge
pandas 1.3.1 py38h1abd341_0 conda-forge
parallel 20160622 1 bioconda
parso 0.8.2 pyhd8ed1ab_0 conda-forge
patsy 0.5.1 py_0 conda-forge
perl 5.32.1 0_h7f98852_perl5 conda-forge
perl-threaded 5.26.0 0 bioconda
pexpect 4.8.0 pyh9f0ad1d_2 conda-forge
pickleshare 0.7.5 py_1003 conda-forge
pillow 8.3.1 py38h8e6f84c_0 conda-forge
pip 21.2.2 pyhd8ed1ab_0 conda-forge
pkg-config 0.29.2 h36c2ea0_1008 conda-forge
pluggy 0.13.1 py38h578d9bd_4 conda-forge
prompt-toolkit 3.0.19 pyha770c72_0 conda-forge
ptyprocess 0.7.0 pyhd3deb0d_0 conda-forge
py 1.10.0 pyhd3deb0d_0 conda-forge
pycparser 2.20 pyh9f0ad1d_2 conda-forge
pygments 2.9.0 pyhd8ed1ab_0 conda-forge
pynndescent 0.5.4 pyh6c4a22f_0 conda-forge
pyopenssl 20.0.1 pyhd8ed1ab_0 conda-forge
pyparsing 2.4.7 pyh9f0ad1d_0 conda-forge
pysam 0.16.0.1 py38hbdc2ae9_1 bioconda
pysocks 1.7.1 py38h578d9bd_3 conda-forge
pytest 6.2.4 py38h578d9bd_0 conda-forge
python 3.8.5 h4d41432_2_cpython conda-forge
python-dateutil 2.8.2 pyhd8ed1ab_0 conda-forge
python_abi 3.8 2_cp38 conda-forge
pytz 2021.1 pyhd8ed1ab_0 conda-forge
readline 8.0 h46ee950_1 conda-forge
requests 2.26.0 pyhd8ed1ab_0 conda-forge
rosella 0.3.3 h443a992_0 bioconda
samtools 1.9 h10a08f8_12 bioconda
scikit-bio 0.5.6 py38h0b5ebd8_4 conda-forge
scikit-learn 0.24.2 py38hdc147b9_0 conda-forge
scipy 1.7.1 py38h56a6a73_0 conda-forge
seaborn 0.11.1 hd8ed1ab_1 conda-forge
seaborn-base 0.11.1 pyhd8ed1ab_1 conda-forge
setuptools 49.6.0 py38h578d9bd_3 conda-forge
six 1.16.0 pyh6c4a22f_0 conda-forge
sqlite 3.32.3 hcee41ef_1 conda-forge
starcode 1.4 h779adbc_1 bioconda
statsmodels 0.12.2 py38h5c078b8_0 conda-forge
tbb 2020.2 h4bd325d_4 conda-forge
threadpoolctl 2.2.0 pyh8a188c0_0 conda-forge
tk 8.6.10 h21135ba_1 conda-forge
toml 0.10.2 pyhd8ed1ab_0 conda-forge
tornado 6.1 py38h497a2fe_1 conda-forge
traitlets 5.0.5 py_0 conda-forge
umap-learn 0.5.1 py38h578d9bd_1 conda-forge
urllib3 1.26.6 pyhd8ed1ab_0 conda-forge
vt 2015.11.10 he941832_3 bioconda
wcwidth 0.2.5 pyh9f0ad1d_2 conda-forge
wheel 0.36.2 pyhd3deb0d_0 conda-forge
xz 5.2.5 h516909a_1 conda-forge
zlib 1.2.11 h516909a_1010 conda-forge
zstd 1.5.0 ha95c52a_0 conda-forge
```

Start by reproducing the command-line script failure, contrasting it with the interactive Python-terminal case and the reported threading-layer or ProcessPoolExecutor settings. Inspect the stack-trace entry points in umap/umap_.py, especially fuzzy_simplicial_set and nearest_neighbors, alongside pynndescent_.py. Done means identifying the cause and preventing the segmentation fault in the reported script execution paths.

Written by the indexing model from the issue text.