Fix flaky CI checks on Ubuntu

[mhucka]

The current pytest runs on Ubuntu often fail during PR checks. It seems that sometimes the optimization steps take too long and time out:

=========================== short test summary info ============================
FAILED cirq-core/cirq/experiments/z_phase_calibration_test.py::test_calibrate_z_phases[angles6-error6-characterization_flags6] - RuntimeError: Optimal parameters not found: The maximum number of function evaluations is exceeded.
= 1 failed, 19825 passed, 121 skipped, 89 xfailed, 7 xpassed, 843 warnings in 334.79s (0:05:34) =
Error: Process completed with exit code 1.

I'm seeing this kind of thing happen on more than 50% of my PRs.

[mhucka]

This seems to be coming from calls to SciPy's curve_fit in the file xeb_fitting.py. The usual reason is that the optimization algorithm failed to find the best fit parameters within the allowed number of iterations.

[mhucka]

What's frustrating about this is that I can't seem to get it to happen locally on a linux system, even if I explicitly set the values of the parameters angles and error in test_calibrate_z_phases_no_options to the same values as those of an example failure on GitHub.

[mhucka]

I've finally been able to reproduce it. I had to set the random seed to the same value as one of the examples of a failure during a CI run.

[mhucka]

This will be fixed by PR #6906 once it's accepted.