bug - overflow in tdcpy.stability.gamma_r.gamma_normalized_diff

[adriansal0712]

1. See the branch bugfix/gamma_diff_overflow
2. Please run the test in tests/test_spectral_abscissa_diff.

There are multiple issues here:

1. compressed diff: in the function spectral_abscissa_diff, it compresses the delays. I think this should not be there if we are discussing the closed_loop diff. If we compress the diff, then the size of the normalized_diff is reduced and it does not take into account the uncertainty of the 0-delay term in the closed-loop.

2. tdcpy.spectral_abscissa.spectral_abscissa_diff: gives incorrect result for case 1

3. overflow: Added a logger.debug message at line 363 of gamma_normalized_diff. The problem occurs in line 359, 365.

To-do: remove the line in code and delete test after bug fix.