Weak hashset adjustment

[jmid]

I just saw another occurrence failing to trigger a parallel STM Weak HashSet failure, this time under 5.1 bytecode:
https://github.com/ocaml-multicore/multicoretests/actions/runs/5488638494/jobs/10001760309

random seed: 192914563
generated error fail pass / total     time test name

[ ]    0    0    0    0 / 1000     0.0s STM Weak HashSet test sequential
[ ]    0    0    0    0 / 1000     0.0s STM Weak HashSet test sequential (generating)
[✓] 1000    0    0 1000 / 1000     2.7s STM Weak HashSet test sequential
================================================================================
success (ran 1 tests)
generated error fail pass / total     time test name

[ ]    0    0    0    0 / 2000     0.0s STM Weak HashSet test parallel
[ ]  742    0    0  742 / 2000    57.3s STM Weak HashSet test parallel
[ ] 1497    0    0 1497 / 2000   117.6s STM Weak HashSet test parallel
[✗] 2000    0    0 2000 / 2000   160.6s STM Weak HashSet test parallel

I therefore ran a bit of failure rate stats only to learn that the error rate is pretty slim under both native and bytecode:

jmi@MintPad:~/software/multicoretests-latest$ dune exec src/weak/stm_tests_hashset_stats.exe -- -v
Weak Hashset 5 / 10000              
jmi@MintPad:~/software/multicoretests-latest$ dune exec src/weak/stm_tests_hashset_stats.bc -- -v
Weak Hashset 5 / 10000              

This PR therefore raises the count from 2000 to 5000.
This decreases the chance of completing a non-error-triggering run from 37% to 0.8% if my calculations are right
$( 1 - 1/2000 ) ^{count}$.

[jmid]

CI summary

• a Windows bytecode 5.1 crash in threadomain [ocaml5-issue] Windows failures on threadomain #203
• a Windows bytecode trunk timeout in threadomain [ocaml5-issue] Windows failures on threadomain #203
• macOS trunk failure to trigger STM int64 ref test parallel asymmetric 'STM _ ref test parallel asymmetric' failure to trigger #364
• 32-bit trunk timeout during excessive shrinking 11841.6s Lin DSL In_channel test with Domain Long In_channel shrink runs causing timeout #373
• Cygwin 5.1 part1 timeout during excessive In_channel shrink run 9035.6s Lin DSL In_channel test with Domain Long In_channel shrink runs causing timeout #373
• Cygwin trunk part1 timeout due to general slowness (multiple +1000s test runs)
• Cygwin trunk part2 timeout due to an excessive Thread ref shrink run: 10251.7s Lin ref int64 test with Thread
• s390x 5.1 bytecode timeout during an excessive Bigarray shrink run 5887.7s Lin DSL Bigarray.Array1 (of ints) test with Domain
• s390x trunk bytecode timeout due to general slowness (multiple +1000s test runs)

None of these are related to the adjusted Weak hashset test, so I'll merge