Increase stress test of threaded gr_mpoly_divides

[oscarbenjamin]

See #2819 (comment)

This changes the parametrisation of the test for heaped gr_mpoly_divides. With these changes the test fails about 50% of the time here on Mac M3 with multiplier 1:

$ FLINT_TEST_MULTIPLIER=1 build/gr_mpoly/test/main gr_mpoly_divides_heap_threaded
gr_mpoly_divides_heap_threaded...
gr_mpoly_divides_heap_threaded                    0.29   (PASS)
$ FLINT_TEST_MULTIPLIER=1 build/gr_mpoly/test/main gr_mpoly_divides_heap_threaded
gr_mpoly_divides_heap_threaded...
FAIL: exact division incorrectly reported as GR_DOMAIN
i = 168, j = 0, dstatus1 = 0, dstatus2 = 1
Integers mod 390041974493255551 (_gr_nmod)
zsh: abort      FLINT_TEST_MULTIPLIER=1 build/gr_mpoly/test/main