set ecc to zero for failures t/f > tmax/fmax

[md-arif-shaikh]

Addressing #206. For methods like Amplitude or Frequency, it could be challenging to detect the small oscillations in the later part of the inspiral and therefore the maximum allowed time tmax or frequency fmax could be much smaller than the time/frequency near merger. Therefore, for time/frequency after tmax/fmax, eccentricity can not be measured. If set_failures_to_zero is True, we set the eccentricity/mean anomaly to zero for such cases at these times/frequencies.