Estim magnetic field is always NaN

[djdupre]

The mag_eci_unit output is consistently NaN for all axes. Taylor told this can happen in env_estim when the TLE year falls outside a valid range, but I'm now using a 2018 TLE (copied from the SWISSCUBE TLE in the uw_hs1_adcs repo) and it's still NaN.

This issue is meant to replace the "NaN BAD" sticky note.