I converted the MICA times for the Spring Equinox between 1800 and 2050 to Julian Dates and got a linear equation for the fit. A plot of the deviations appears to be fairly random with a maximum deviation of about 16 minutes.
The "resonance" seems to have disappeared.
Supplemental (Apr 7): The probability distribution for the deviations from the linear fit doesn't appear to be normal. Its peak appears to be somewhat flattened.