IDEX failing PROD l2a job.

[lacoak21]

https://dagster.imap-mission.com/runs/d1a88786-38c8-4f8d-8c6b-855a78dd4a26

[lacoak21]

The issue is: In fit_impact when exponent 1 becomes infinite then multiplied by 0 (from np.heaviside).

@aldo9253 over email said:

I think we can replace fit_impact with:

fit_impact():
time_since_impact = time - time_of_impact
elapsed_time = np.maximum(time_since_impact, 0.0)

exponent_1 = 1.0 - np.exp(-elapsed_time / rise_time)
exponent_2 = np.exp(-elapsed_time / discharge_time)

return constant_offset + np.heaviside(time_since_impact, 0) * (
amplitude * exponent_1 * exponent_2
)

This ought to solve the problem with no science impact.