More stable run_from_experiment test

[Ericgig]

Description
The test for the new method run_from_experiment has a time step too long for the euler method. This would cause fail once a few hundreds time as in https://github.com/qutip/qutip/actions/runs/8979789558/job/24662362047.
Reducing the system size with smaller steps makes the test more consistent (no failure in 20000 tries.)

Also reduced test_feedback step size to improve that test consistency.

[coveralls]

coverage: 86.09%. first build
when pulling 8304a3a on Ericgig:misc.better_stoc_test
into 94f7392 on qutip:master.