ZTF Alert FP - deduplication logic, flux to mag space

[Theodlz]

This PR:

• add the deduplication logic folks decided to use, i.e. when we have overlapping FP datapoints, replace existing ones only if the new alert is brighter (which means, the position should be more accurate).
• add mag space information when the SNR is above 0. That way, filters that need the mag can trust it or not trust it by enforcing their own SNR cuts.
• ensure that only brand new ZTF objects start accumulating forced photometry. That way, we either don't have FP, or we have the full history for that object. Otherwise, filters relying on FP could yield unpredictable results.
• add tests for the deduplication

TODO: add tests for the mag space data maybe?

[mcoughlin]

a few thoughts / minor doc edits.

[Theodlz]

Thanks a lot for the review @mcoughlin (and sorry for the stupid english mistakes, you make one of them and then copilot propagates that error as much as possible yikes).