Signal priors

[tdixon97]

Closes #9 , for now sqrt(S) is a bit hardcoded, in future would be better to make it a distribution object (like other priors)

[sofia-calgaro]

i fear including sqrt_prior in the likelihood is not the best Like what if we want to include a 3rd prior right now? do we have to add other 2 terms as input for the current likelihood? do we rename the present sqrt_prior=false,s_max=nothing?

in principle we could build a dictionary of priors from where to retrieve necessary info, not sure though if we need sth more than s_nax for different priors. At that point, one idea would be to create one likelihood per prior

[tdixon97]

for sure its not the cleanest solution, i think the other solution is to define a custom distribution but thats not so easy. For sure we could think about passing around the info in a cleaner way than having so many arguments to functions. For example we could create a 'fit_info' struct with all the infos Btw regarding other priors yeah some will also need an Smin since they arent normalisable (eg log(S).

[sofia-calgaro]

should I close it to test it and then we can think of a better solution in a separate issue? changes are not so huge to come here another time in the future

[tdixon97]

i would close it, since in the end we dont need the 'perfect' solution since in practice we only need a few prior choices