New det_cal db introduces NaNs in timeconst_filter

[mmccrackan]

For a new site pipeline run of satp1 data, I updated the manifest directory to point to a more recent det_cal db (switched from satp1_det_cal_240312m to satp1_det_cal_240910m). It seems that this introduced NaNs into the axis_manager signal during the
timeconst_filter Fourier filter stage. Using the older det_cal does not produce this issue. @jlashner and @msilvafe.

An example observation where this occurs is:
obsid='obs_1717988397_satp1_1111111'
dets={'wafer_slot': 'ws3', 'wafer.bandpass': 'f150'}

The config filter params are:

- name: "fourier_filter"
      process:
        filt_function: "timeconst_filter"
        filter_params:
          timeconst: "det_cal.tau_eff"
          invert: True