Implement a detection model

[mi-dai]

Would like to implement a detection model that can flag transients as detected or not (and make a decision on whether to write out the light curve if they are not detectable). We may be able to make some estimation on the maximum S/N and filter out things that are definitely not detectable to save some compute time in calculating the full light curve.

For LSST alerts, I think this is something like an observation with S/N > 5 (or some other number);
For Platicc, I believe it's 2 observations with S/N >5 (to filter out moving objects)

We may want to write this as a function/module with required number of observations and S/N as parameters.