initial commit of simple rewriting of certain components of AssayTool…

[yuanqing-wang]

…s in NumPyro. targetting replicating functionality displaced in direct-fluorescence-assay

[alexjli]

I think inference is still a bit unstable, which I'm working on. Attached above is one of the runs where the parameters are a little off (in particular, f_measure_var models measurement error in fluorescence with a true value of 10, while this inference run ends with a final value of about 0.5). The predictive model generally models the data pretty well regardless though.