Disable catch_signals in sc_init

[benegee]

As found in #35, it might help with current CI failures.

[spenke91]

Thanks!

[benegee]

@spenke91 and I noticed that running with a single thread explicitly avoids the current segmentation fault issues in CI.
We did this via environment variables, see

T8code.jl/.github/workflows/test.yml

Line 49 in 5a77ba3

JULIA_NUM_THREADS: 1

, and via command line for a newly spawned julia process, see

T8code.jl/test/runtests.jl

Line 29 in 0351fab

run(`$(Base.julia_cmd()) --threads=1 --check-bounds=yes --project=$(dirname(@__DIR__)) $(abspath("test_all.jl"))`)

.

I still do not understand why this changes anything. At least I verified locally that nthreads and ngcthreads return 1 each, regardless of the threads setting.

Anyway, this suggests #35 could be helpful here. According to https://github.com/cburstedde/libsc/blob/f83ea609a019178af2742f11e1a300c8ee78e9d7/src/sc.h#L813, it disables catching of SIGINT and SIGSEGV.